|Home | About | Journals | Submit | Contact Us | Français|
Vaccinia virus (VV) is an enveloped DNA virus from the poxvirus family and has played a crucial role in the eradication of smallpox. It continues to be used in immunotherapy for the prevention of infectious diseases and treatment of cancer. However, the mechanisms of poxvirus entry, the host factors that affect viral virulence, and the reasons for its natural tropism for tumor cells are incompletely understood. By studying the effect of hypoxia on VV infection, we found that vascular endothelial growth factor A (VEGF-A) augments oncolytic VV cytotoxicity. VEGF derived from tumor cells acts to increase VV internalization, resulting in increased replication and cytotoxicity in an AKT-dependent manner in both tumor cells and normal respiratory epithelial cells. Overexpression of VEGF also enhances VV infection within tumor tissue in vivo after systemic delivery. These results highlight the importance of VEGF expression in VV infection and have potential implications for the design of new strategies to prevent poxvirus infection and the development of future generations of oncolytic VV in combination with conventional or biological therapies.
Vaccinia virus (VV), the prototypic and most extensively characterized member of the Orthopoxvirus genus of the Poxviridae, has played a crucial role in one of the greatest achievements in medicine—the eradication of smallpox (1)—and continues to be used as an immunotherapeutic approach for the prevention of infectious diseases and treatment of cancer (2–4). VV has several inherent features that make it particularly suitable for use as an oncolytic agent (3) including fast and efficient replication, rapid cell-to-cell spread, natural tumor tropism, strong lytic ability, large cloning capacity (>25 kbp), well-defined molecular biology, safety in human beings, and good stability. A defining feature of VV as an oncolytic agent is that it relies on its own encoded proteins to carry out replication and transcription in the cytoplasm. Only a small number of cellular proteins are involved in VV infection (5). This may allow VV to replicate in many different cell types and overcome the serious problem of more limited tropism encountered with adenovirus, another, more commonly used oncolytic vector (3, 6). Oncolytic VV has demonstrated benefit in recent phase I/II clinical trials (7, 8). These trials have indicated the inherent safety of the virus, efficacy of intravenous delivery of VV, VV infection of metastatic deposits, and reproducible disease control in many patients enrolled in the trials. However, along with other groups (9), we have observed that there is variability in cell killing by VV in different cancer cell lines, and the reasons for this are not fully understood (10).
Previously it has been shown that a hypoxic microenvironment, commonly found in solid tumors, is detrimental to the replication and efficacy of many types of oncolytic viruses (11–13). However, we have demonstrated recently that in contrast to other virus species, the cytotoxicity of oncolytic VV can actually be enhanced by hypoxia in some cell lines (14). This finding is an additional biological property that makes VV an ideal vector for targeting solid tumors. Further investigation of this property and the interaction of tumor cells and VV may offer insight into VV biology and aid the development of new strategies for both cancer therapy and the prevention of Poxvirus infection.
Here we report that this hypoxic induction of viral cytotoxicity was found only in those cell lines with a concordant hypoxic induction of vascular endothelial growth factor A (VEGF-A) expression. Functional studies using small interfering RNA (siRNA) gene silencing and stable overexpression of VEGF-A show that VEGF-A can augment viral transgene expression and replication in vitro and in vivo in both human and murine models. Dissection of the viral life cycle demonstrated that VEGF-A, via Akt activation, facilitates the internalization of both wild-type VV and recombinant VVL15 (thymidine kinase [TK]-deleted virus expressing firefly luciferase) and is an important cellular factor affecting the tropism of VV for tumor cells.
The human pancreatic carcinoma cell lines SUIT-2, CFPac1, MiaPaca2, Panc1, PaTu8988t, and PaTu8988s were obtained from Cancer Research UK Central Cell Services (CRUK CCS, Clare Hall, Herts, United Kingdom) and maintained in Dulbecco's modified Eagle medium (DMEM) with 10% fetal calf serum (FCS) and supplemented with 0.06 μg/liter penicillin and 0.1 μg/liter streptomycin. Normal human bronchial epithelial (NHBE) cells (Lonza) were maintained in bronchial epithelial growth medium (BEGM). Cell lines were maintained in their respective media at 37°C under normoxic (20% O2 supplemented with 5% CO2) or hypoxic (1% O2 supplemented with 5% CO2) conditions as indicated.
The wild-type Lister vaccine strain of VV and recombinant thymidine kinase (TK)-deleted VV (VVL15) were a gift from Istvan Fodor (Loma Linda University Campus, California). These were produced as previously described (15) and propagated in CV1 (green monkey kidney) cells. The fluorescently tagged VVL-488 was produced by labeling wild-type VV with Alexa Fluor 488 5-sulfodicholorphenol ester (Invitrogen) as previously described (16).
The VEGF-A p165 isoform transcript (NM_001025368.1) was cloned into the pCMV6-Neo eukaryotic expression vector (Origene). MiaPaca2 cells were transfected with the VEGF-A p165 plasmid or the empty expression vector using Effectene (Qiagen) according to the manufacturer's instructions, and stable cell lines were selected using 1 mg/ml neomycin. These cell lines were designated MPVe-165 (expressing VEGF-A) and MPVC (transfected with the empty expression vector). For all experiments using MPVe-165, a minimum of two clones of the stable cell line were tested to ensure valid results, and in no case was a significant difference in the behavior of each stable cell line clone observed.
To silence VEGF-A gene expression, SUIT-2 cells were transfected with 25 nM SmartPool VEGF-A siRNA or SiGenome Risc-free Control siRNA (Dharmacom) using the Dharmafect transfection reagent. All viral assays were performed 72 h after siRNA transfection in serum-free media at the point of maximal VEGF-A gene silencing.
VEGF-A protein levels were quantified using a VEGF-A-specific enzyme-linked immunosorbent assay (ELISA) (R&D Systems) according to the manufacturer's instructions. Experiments were performed in duplicate, and quantification was performed in triplicate.
Cells were seeded in triplicate and infected 16 h later with wild-type VV. Cells and supernatant were harvested and freeze-thawed three times. Titers were determined by measuring the 50% tissue culture infective dose (TCID50) on indicator CV1 cells. The cytopathic effect was determined by light microscopy 10 days after infection. The Reed-Muench mathematical method was used to calculate the TCID50 value for each sample (17). Triplicates were used for each time point, and each replicate was assayed twice for cytopathic effect. Viral burst titers were converted to PFU per cell based on the number of cells present at viral infection.
The cytotoxicity of the virus was assessed 6 days postinfection (p.i.) with virus using an MTS nonradioactive cell proliferation assay kit (Promega) according to the manufacturer's instructions. Cell viability was determined by measuring absorbance at 490 nm using a 96-well plate absorbance reader (Dynex), and a dose-response curve was created by nonlinear regression, allowing determination of a 50% effective concentration (EC50) (viral dose required to kill 50% of the cells). Each assay contained six replicates, and each assay was repeated four times.
Cells were infected with VVL15 and bioluminescence measured using an IVIS bioluminescent imager (Xenogen). Cells were seeded in triplicate and infected 16 h later. At each time point, luciferase expression was determined according to the manufacturer's instructions. In brief, medium was aspirated and replaced with 150 μg/ml d-luciferin (Xenogen Corp., California) in phosphate-buffered saline (PBS) at 37°C, and luminescence was measured after 2 min. Light emission was quantified as the sum of all detected photon counts within uniformly sized regions of interest (ROIs) with each well manually defined during post-data acquisition image analysis. This was measured in photons per second per cm2 and expressed as relative light units (RLU) for the purpose of this paper.
All steps were carried out with PBS buffer containing 1% bovine serum albumin (BSA). Appropriate cells were suspended (1.5 × 105 in 100 μl) in test tubes and incubated with 100 μl of buffer alone or buffer containing wild-type VV (multiplicity of infection [MOI], 10 PFU/cell) at 4°C with vigorous shaking for 1 h. Cells were washed three times with cold PBS-BSA to remove unbound virus particles and placed at 37°C for 15 or 30 min (subsequently referred to as 15 or 30 min p.i.). The attached but uninternalized viral particles were removed with 1 mg/ml pronase, and the samples were collected for DNA preparation. Pronase has been reported to reduce more than 90% of intracellular mature virus (IMV) binding to cells if the cells have been pretreated with it (18). In this study, we first confirmed that pronase at 1 mg/ml could also effectively remove more than 75% of attached VV particles from the cell surface after 30 min of treatment (data not shown). The viral genome copy number was determined using TaqMan quantitative real-time PCR (qPCR) with VV late transcription factor 1 (VLTF-1) primers and probes as previously described (19). The primers were S′, AACCATAGAAGCCAACGAATCC, and AS′, TGAGACATACAAGGGTGGTGAAGT, and the probe sequence was ATTTTAGAACAGAAATACCC. Forty nanograms of total DNA was used as the template. The viral DNA copy number was normalized by 40 ng of total DNA and presented as viral DNA copy number/ng of total DNA. All experiments were repeated a minimum of three times.
Monolayers of SUIT-2 cells transfected with VEGF-A or control siRNA were grown in chamber slides (10,000 cells/well). Slides were incubated at 4°C for 1 hour before infection with VVL-488, and the medium was replaced with fresh infection medium containing VVL-488 and incubated at 4°C for 1 hour to allow the virus to bind to the cells. After 1 hour of cold incubation, the cells were washed twice with cold PBS, incubated in warm media at 37°C for 30 min, washed twice with prewarmed PBS, fixed in methanol at −20°C for 10 min, and then blocked for 1 h with PBS (1% BSA). Alpha-tubulin (Sigma-Aldrich) antibody was used at a 1:500 dilution. DNA was stained with 4′,6-diamidino-2-phenylindole (DAPI). Confocal images were taken with a 63× lens on a Zeiss confocal microscope. This technique has been used by our group and others to track viral entry (16, 19).
SUIT-2 cells were seeded at 4 × 105 cells per well in DMEM plus 10% FBS in 6-well plates and left to adhere overnight. The following day, medium was removed and replaced with serum-free medium overnight. Cells were then treated with recombinant human (rh) VEGF-165 (10 ng/ml) (Peprotech), and lysates were harvested at 0 min (no VEGF treatment), 3, 5, 10, 20, and 30 min following VEGF addition, using phospho-safe lysis buffer (Merck). Total and phospho-AKT levels (S473) were analyzed by Western blotting using anti-rabbit tAKT or anti-rabbit P-AKT (S473) antibodies (Cell Signaling). Protein levels were quantified using ImageJ Software.
Normal human bronchial epithelial (NHBE) cells were seeded at 2 × 105 cells per well in 6-well plates using complete BEGM (Lonza) and left to adhere overnight. The following day, once the cells had reached 80% confluence, medium was replaced with BEGM minus bovine pituitary extract (BPE) and epidermal growth factor, and cells were starved overnight. Akt inhibitor VIII (Calbiochem) was added to the cells (20 μM) for 2 h. After 2 h, rhVEGF-165 (Peprotech) was added (10 ng/ml) to each well, and the cells were incubated for a further 10 min before infection with wild-type VV at 0.1 PFU/cell. Cells were then collected by scraping every 24 h for 72 h. Cell lysates were freeze-thawed three times before TCID50 was determined as previously described.
The VEGF receptor (VEGFR) inhibitor pazopanib (GW-786034; Biovision) was used at a final concentration of 30 nM to prevent VEGFR-mediated cell signaling. Briefly, MPVe-165 or MPVC cells were seeded and serum starved overnight as previously described. Pazopanib was added to each well for 30 min in serum-free medium. Cells were then infected with VVL15 and used for downstream assays.
Animal studies were performed according to the guidelines for the welfare and use of animals in cancer research (20) and in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986. MPVC and MPVe-165 xenografts were established in the right flank of BALB/c nude mice (5 per group) by injecting 5 × 106 cells. When tumors reached 0.4 to 0.5 cm in diameter 10 to 14 days later, mice received 100-μl intravenous tail vein injections of 1.0 × 107 PFU of VVL15. The biodistribution of VVL15 was determined in anesthetized mice (2% isoflurane inhalation) after intraperitoneal injection of d-luciferin (150 mg/kg of body weight) (Xenogen), and fluorescence was measured with the IVIS camera (Xenogen Corp.) in tumors defined as regions of interest after imaging.
Immunohistochemistry was performed on 4-μm-thick frozen sections of tissues with rabbit anti-VV coat protein polyclonal antibody (MorphoSys UK Ltd.) or Pecam-1 (ab56299; Abcam Ltd.). The Ventana Molecular Discovery System was used for staining of slides. Images were taken on a Zeiss Axioplan microscope, and all image analysis was performed using Image J software (NIH) after montage reconstruction of whole sections.
The unpaired Student t test, one-way analysis of variance (ANOVA), and two-way ANOVA were used as indicated.
We have previously demonstrated that hypoxia increases the cytotoxicity of oncolytic VV in one-half of the pancreatic ductal adenocarcinoma (PDAC) cell lines screened (14). VEGF-A, one of the target genes strongly induced in hypoxia (21), was one of the candidate host factors that we hypothesized may play a role in modulating the VV life cycle. We cultured four human PDAC cell lines under normoxic (20% O2) or hypoxic (1% O2) conditions and assessed the supernatant for VEGF-A secretion after 48 h, using ELISA. A statistically significant induction of VEGF-A after exposure to 1% oxygen was seen only in the MiaPaca2 and CFPac1 cell lines (Fig. 1A). Interestingly, our previous findings showed that VV was more cytotoxic to these cell lines in hypoxia but not to SUIT-2 or Panc1 cell lines, neither of which showed any significant hypoxic induction of VEGF-A (Fig. 1A). To determine whether the relationship between VEGF expression and sensitivity of cancer cells to VV exists under normoxic conditions, VEGF expression and viral replication of VV in six human pancreatic cancer cell lines were investigated by ELISA and TCID50 assay, respectively (Fig. 1B). The VV replication in the cancer cell lines correlated with the level of VEGF expression under normoxic conditions (R2 = 0.77, P = 0.02).
To assess the functional significance of VEGF-A expression for VV infection, we investigated whether stable overexpression of VEGF-A in cancer cell lines could alter VV infection. MiaPaca2 cells, which have a low endogenous level of VEGF-A, were stably transfected to express the predominant VEGF-A isoform p165 (21). The resultant stable cell line, designated MPVe-165, had an approximately 2-fold increase in VEGF expression in comparison to its paired vector-transfected control cell line (MPVC), and this increase was maintained over the course of the experiment (Fig. 2A). To assess the effect of VEGF-A on VV gene expression, MPVe-165 and MPVC were infected with VVL15 (TK-deleted VV expressing the firefly luciferase reporter gene) at an MOI of 1 PFU/cell. The overexpression of VEGF-A resulted in a >3-fold increase in viral gene expression at 4 h postinfection and >2-fold increase at 24 h postinfection (Fig. 2B) as determined by quantification of bioluminescence arising from expression of the luciferase reporter gene. To verify the effect of VEGF-A overexpression on the production of infectious virions, MPVe-165 and MPVC were infected with wild-type VV strain Lister at an MOI of 1 PFU/cell. Lysates were collected at 24, 48, and 72 h postinfection, and CV1 cells were used as indicator cells to determine a viral titer using a TCID50 assay. A statistically significant increase in viral replication was seen in MPVe-165 cells at 48 and 72 h postinfection compared to MPVC (Fig. 2C).
VEGF-A also increases the oncolytic potential of VV. MPVe-165 and MPVC cells were seeded in 96-well plates and infected with serial dilutions of wild-type Lister strain virus. A cell cytotoxicity assay (MTS; Promega) was performed at 6 days postinfection to determine the percentage of viable cells remaining, and a dose-response curve was drawn in order to calculate the EC50 (Fig. 2D). A statistically significant reduction in EC50, which implies an increase in viral cytotoxicity, was seen in MPVe-165 compared to MPVC cells.
In order to validate further the effect of VEGF-A on VV infection, we investigated whether downregulation of VEGF-A by siRNA could reduce VV infection in tumor cells. Given that SUIT-2 has a high endogenous level of VEGF-A expression compared to MiaPaca2 (Fig. 1), this cell line was chosen to evaluate the effects of VEGF-A-specific siRNA on VV infection. After optimization, we confirmed that VEGF-A SmartPool siRNAs dramatically suppressed VEGF-A expression in SUIT-2 cells 48 h after transfection, and this lasted for at least 96 h (Fig. 3A). The control siRNA did not affect cell viability or apoptosis induction at the concentration used in the present study. In contrast to the overexpression of VEGF-A, downregulation of VEGF-A expression significantly reduced reporter gene (luciferase) expression (Fig. 3B). We found that SUIT-2 cells treated with VEGF-A siRNA had a 10-fold reduction in bioluminescence in comparison to control siRNA-treated cells, and this was consistent at different MOIs. We also found a statistically significant reduction in infectious virions produced in SUIT-2 cells after VEGF-A gene silencing at 48 h (59.6 versus 156.2 PFU/cell), 72 h (302.2 versus 553.8 PFU/cell), and 96 h (168.2 versus 344.9 PFU/cell) postinfection (Fig. 3C). Downregulation of VEGF-A expression also decreased the cytotoxicity of VV (Fig. 3D), with an increased EC50 (2.2 PFU/cell in SUIT-2 cells transfected with VEGF-A siRNA compared to 1.8 PFU/cell in SUIT-2 cells transfected with control siRNA). Furthermore, the attenuated viral replication in SUIT-2 cells after transfection with VEGF-siRNA could be rescued by adding rhVEGF protein (Fig. 3E).
To confirm further the importance of VEGF-A-mediated signaling for VV gene expression and replication, MPVe-165 and MPVC cells were treated with 30 nM pazopanib (GW-786034), a clinically relevant inhibitor of VEGF receptors (VEGFR), in particular,VEGFR1, VEGFR2, and VEGFR3 (Biovision), 30 min prior to infection with VVL15. By measuring bioluminescence 24 h postinfection, we found that pretreatment of cells with pazopanib significantly reduced reporter gene expression (Fig. 3F) and replication (Fig. 3G) in the engineered MiaPaca2 cells. This result not only demonstrates that VEGF-A mediated signaling dramatically improves VV gene expression and replication but also confirms the previously reported autocrine nature of VEGF-A-mediated signaling (22).
Since VEGF-A expression affects the induction of early gene expression at 4 h prior to any viral replication (Fig. 2B), we hypothesized that VEGF-A might affect phases of VV infection between attachment and early gene expression. We therefore assessed the effect of VEGF-A on VV attachment and internalization in tumor cells using quantitative PCR. MPVe-165 and MPVC were infected with wild-type VV, and although there was no difference in attachment of VV (Fig. 4A), there was a significant increase in viral internalization at 30 min postinfection (Fig. 4B). Similarly, we found no difference in viral attachment to SUIT-2 cells treated with VEGF-A siRNA versus control cells but a significant reduction in internalization at 30 min postinfection in VEGF-A siRNA-transfected SUIT-2 cells (Fig. 4C). We further confirmed this finding using fluorescent confocal microscopy and a fluorophore-labeled wild-type VV (VVL488). SUIT-2 cells were treated with VEGF-A or control siRNA and infected with VVL488, and images were then taken at infection and 30 min postinfection. Quantification of attachment and internalization was performed using z-stack three-dimensional (3D) representations of 10 fields of view for the VEGF siRNA- and control siRNA-treated SUIT-2 cells. A clearly isolated green fluorescent protein (GFP) focus was counted as one virion, and small and large foci counted equally. Two independent observers, who were blinded to sample groups, scored images at the time of viral attachment and 30 min postinfection. No difference was seen in viral attachment, but a significant (P = 0.023) reduction in internalization was seen in VEGF-A siRNA-treated SUIT-2 cells versus control (1.5 versus 2.4 foci per cell, respectively). Representative images of cells at 30 min postinfection are shown in Fig. 4D and andE.E. To test whether VEGF affects other steps of life cycle of VV after internalization, the effects of VEGF-A on viral mRNA production, viral DNA replication, and the egress of extracellular enveloped virus particles were assessed. After accounting for the increased internalization of VV due to VEGF-A, we found no other effect on later stages of virus infection (data not shown).
The standard models of VEGF-A biology focus on tumor cell secretion of VEGF-A acting on endothelial cells to stimulate neoangiogenesis (21). One of the functions of VEGF-A on endothelial cells is to stimulate, via activation of Akt, remodeling of the actin cytoskeleton to facilitate cell motility (23, 24). VV internalization is known to be dependent on actin remodeling at the cell membrane (25), so we hypothesized that VEGF-A may be acting to activate Akt and facilitate VV entry.
To determine the effect of VEGF-A stimulation on Akt activation, SUIT-2 cells were serum starved and then stimulated with 10 ng/ml recombinant human VEGF-A (rhVEGF) (Peprotech) or 10 ng/ml BSA in fresh medium. Immunoblotting demonstrated a transient increase in the phosphorylated (activated) form of Akt (S473) at 3 and 5 min after rhVEGF stimulation (Fig. 5A), while BSA did not induce a significant alteration of phosphorylated Akt over the time course (Fig. 5A, lower panel). This increase was prevented by pretreatment of the cells with the VEGFR inhibitor pazopanib (Fig. 5B), showing that VEGF-A can, by signaling through the VEGFR, activate intracellular Akt.
Having demonstrated that VEGF increases Akt activation, we went on to test whether Akt inhibition would affect VV internalization. We used Akt Inhibitor VIII, a selective, reversible Akt inhibitor that blocks PH (pleckstrin homology) and kinase domains interactions. This restricts phosphorylation of Akt at Thr308 and Ser473, reduces the levels of active Akt in cells, and blocks the phosphorylation of known Akt substrates (26). SUIT-2 cells were treated with Akt Inhibitor VIII at 5 μM or 20 μM or with dimethyl sulfoxide (DMSO) prior to infection with wild-type VV, and a standard internalization assay was performed as detailed in Materials and Methods (Fig. 5C). There was a significant reduction in VV internalization at 15 min postinfection with 20 μM (the highest dose) inhibitor (47% reduction versus reduction in DMSO; P = 0.009) and a trend toward reduction with 5 μM (P = 0.07). At 30 min postinfection, there were reductions of 30% and 35% in VV internalization after treatment with Akt inhibitor VIII at 5 μM (P = 0.042) and 20 μM (P = 0.032), respectively. There was no detrimental effect on cell viability at 24 and 48 h with transient exposure to the Akt inhibitor at these concentrations.
Poxviruses, including both variola virus and VV, are able to enter via the respiratory epithelium, and since it has been reported that respiratory epithelium expresses both VEGF-A and its receptors (27–30), we were interested to determine whether exogenous VEGF-A could increase the susceptibility of NHBE cells to VV infection. Having confirmed that rhVEGF-A significantly increased reporter gene expression of VV (Fig. 6A), NHBE cells were stimulated with rhVEGF-A following transient exposure to the Akt inhibitor and then infected with wild-type VV (MOI = 0.1 PFU/cell). Viral replication was titrated using a TCID50 assay at 72 h p.i. There was a significant increase in viral replication following stimulation with rhVEGF-A, and this could be blocked by inhibition of Akt phosphorylation prior to viral infection and internalization (Fig. 6B). There was no detrimental effect on uninfected NHBE cell viability with transient exposure to Akt inhibitor at this concentration. This suggests that VEGF facilitates VV infection in both transformed and nontransformed cells. Inhibition of Akt can significantly reduce VV replication in normal epithelial cells.
VEGF-A facilitates VV entry and results in increased gene expression, replication, and cytotoxicity in vitro. To determine if this finding was reproducible in vivo and could be partly responsible for the natural tropism of VV for tumors, we used the MiaPaca2 cell model to establish tumor xenografts. MPVe-165 and control MPVC cell lines were used to establish flank xenografts in BALB/c nude mice. When the MPVe-165 and MPVC tumor xenografts were of a similar size, mice were injected intravenously with 1 × 107 PFU of VVL15, and bioluminescence was quantified. There was an increase in mean bioluminescent signals in MPVe-165 xenografts compared to those of the control group. This was statistically significant at 2.5, 3.5, and 5.5 days p.i., and there was a trend to statistical significance at 1.5 (P = 0.055) and 4.5 (P = 0.067) days p.i. (Fig. 7A). VV infection was confirmed by immunohistochemical staining for VV coat protein, and an example of infected tumor foci is shown in Fig. 7B.
This finding reflects the effect of VEGF-A on VV in vitro, but increased distribution in vivo as a result of increased tumor vascularity consequent to the neoangiogenic effects of VEGF-A is also likely. To confirm this, MPVe-165 and MPVC xenografts were stained for Pecam-1, a marker of murine endothelial cells (31). There was an increase in Pecam-1-positive cells in the MPVe-165 group compared to controls, and representative immunohistochemical images and quantification of the positive staining are shown (Fig. 7C and andDD).
The development of oncolytic viral therapy has recognized the importance of host cell and virus interactions in optimizing therapeutic efficacy. The mechanisms of poxvirus entry, the host factors that affect viral virulence, and the reasons for its natural tropism for tumor cells are incompletely understood. In particular, the mechanism of VV cell entry is opaque. It has been suggested that VV cell entry may occur by passive fusion with the cell membrane, or via uncharacterized receptors, or by macropinocytosis, followed by internalization requiring the coordination of complex actin dynamics (32–34). VVs, including vaccine strains, have been shown to inherently target tumors (10, 35). This tumor selectivity is partially attributed to the activation of the epidermal growth factor/Ras-MEK-extracellular signal-regulated kinase (ERK) pathways and deregulation of interferon (IFN) pathways in cancer cells (36, 37). A greater understanding of the cellular factors essential for efficient VV infection would offer further insights into poxvirus biology and help to improve the development of VV as a novel anticancer therapeutic.
We have previously shown that the cytotoxicity of oncolytic VV is augmented by hypoxia in some PDAC cell lines (14). In this study, we show that this is only in cell lines where there is hypoxic induction of VEGF-A. We used PDAC cell models with both VEGF-A overexpression and siRNA-mediated gene silencing to demonstrate that VEGF-A increases viral reporter gene expression, replication, and cytotoxicity. Quantitative PCR and direct fluorescent confocal microscopy of a fluorophore-labeled virus showed that VEGF-A increases the rate of virion internalization. The use of the clinically relevant inhibitor of the VEFGR (pazopanib) in cells stably expressing VEGF-A also demonstrated that VEGF-A can act in an autocrine manner to increase viral internalization, gene expression, and replication. We show here that VEGF-A can increase Akt phosphorylation, which is prevented by inhibition of the VEGFR, and that selective Akt inhibition reduces VV internalization. Using systemic administration of VVL15, a recombinant VV expressing fLuc, we were able to show that this finding is consistent in vivo. We also demonstrated that this is replicated in NHBE cells, which are a natural host cell for VV.
Although other growth factors such as EGF and its receptor have recently been reported to augment the uptake of adeno-associated virus 6 (AAV6) and influenza virus A (38, 39), we have shown the role of VEGF-A in the VV life cycle. It is interesting that orf virus, from the genus Parapoxvirus, produces a viral virulence factor, VEGF-E, which is the VEGF homologue most closely related to VEGF-A and contains all eight cysteine residues of the central cysteine knot motif characteristic of the VEGF family (40). In addition, myxoma virus, a rabbit-specific poxvirus, has been shown to infect and kill human cancer cells only where there is endogenous Akt phosphorylation and activation (41). Neuropilin-1, a coreceptor of VEGF, is a component of receptor complexes involved in the entry of human T-cell leukemia virus (42).
This is the first study to our knowledge to demonstrate that VEGF-A signaling facilitates VV entry. In our study, silencing VEGF-A or inhibiting Akt activation did not prevent viral infection completely, supporting the concept that there are likely to be multiple pathways by which VV may enter cells, which may differ across cell types (43). Integrin β1-mediated phosphatidylinositol 3-kinase (PI3K)/Akt activation has very recently been reported to be essential for vaccinia virus entry (44). Integrin β1 is important for both vaccinia virus attachment and penetration. It has been reported that VEGFs and their receptors interact with integrins, including β1, on endothelial and tumor cells (45, 46). Further investigations are warranted to dissect whether VEGF interacts with integrin to affect the life cycle of VV and which VEGFR(s) is involved in such effects. However, our present study at least suggests that VEGF-A is one of the cellular factors that are responsible for the tumor tropism of VV.
In addition, an increase in vascularity induced by overexpression of VEGF-A may account for some of the effects in vivo, although it does not give a conclusive indication as to the permeability of these vessels, which may also be important for VV spread (47). Consequently, VEGF-A results in increased tropism of VV for malignant cells by increasing internalization and may also improve virus delivery in vivo due to its role in tumor angiogenesis. The effect of the angiogenic state on tumor uptake of VV, cross traveling of VV from blood to tumor cells, and their relationship with VEGF expression are very interesting topics that warrant further investigations. Of note, our findings have implications for regimens that combine anti-VEGF/Akt agents with oncolytic VV for cancer treatment. It might be more reasonable to use anti-VEGF/Akt agents after treatment with oncolytic VV based on our findings in this study.
VV was widely used in the mass vaccination programs that eradicated smallpox in 1972. Although the immunity of vaccinated individuals lasts for decades (48), a large proportion of the general population is still recognized to be extremely susceptible to variola virus and other poxviruses in the event of an intentional or unintentional release (49). Therefore, it is important to look into more approaches to prevent and or interrupt poxvirus infection, especially in normal human cells. The identification of VEGF-A and Akt signaling as influences in VV infection in normal human epithelial cells (Fig. 6) has important implications for treating poxvirus infection through targeting host cellular genes.
This project was supported by Cancer Research UK (C633-A6253/A6251), Department of Science and Technology, as well as the Department of Health, Henan Province, China (124200510018 and 104300510008).
We declare that we have no conflicts of interest.
Published ahead of print 26 December 2012