The Oncolytic Effect of Recombinant Vesicular Stomatitis Virus Is Enhanced by Expression of the Fusion Cytosine Deaminase/Uracil Phosphoribosyltransferase Suicide Gene

INTRODUCTION

The delivery of suicide genes into malignant cells affords the opportunity of increasing the efficacy of tumor eradication while limiting toxic side effects that are often found with systemic chemotherapy (1) . Suicide gene therapy involves the transduction of cancer cells with a nonmammalian gene encoding an enzyme that converts a relatively innocuous, systemically delivered prodrug into a highly toxic chemotherapeutic (1) . However, a major issue when considering enzyme-prodrug systems for the treatment of cancer remains the effective transport of the suicide gene into the appropriate tumor tissue. Although a number of strategies have been used to target tumor cells effectively, including the use of replication-defective virus vectors, problems associated with low levels of gene transfer and expression are encountered frequently (2) .

Recently, we genetically modified a replication-competent oncolytic virus, VSV, 1 to carry foreign genes such as cytokines (3) . We reasoned that this approach could improve tumor cell killing by combining the effect of direct viral oncolytic activity with the augmented antitumor immune responses or, in the case of suicide genes, bystander cytotoxic action. VSV is an enveloped, negative-stranded RNA virus with a simple genetic structure of five genes and well-characterized immunobiology (4) . VSV is relatively innocuous in humans, inducing at the most a flu-like syndrome (5) . However, the majority of the population lacks antiviral antibodies that could conceivably limit virus treatment, and in general, the incidence of exposure to the VSV is low (6) . VSV is a nonintegrating virus that replicates in the cytoplasm, does not undergo genetic reassortment, and has no known transforming potential (4) . Evidence indicates that the mechanism of VSV-induced oncolysis involves taking advantage of defects in the IFN host defense system prevalent in tumor cells. Lack of an effective IFN response allows propagation of virus, leading to rapid cytolysis (7 , 8) . Our attempts at improving the efficacy of VSV tumor therapy have indicated that rVSV is easy to generate and is amenable to high-titer production and purification (3) . rVSV variants generated thus far have also demonstrated greater oncolytic activity than the WT virus in tumor therapy studies and appear greatly attenuated in vivo compared with the WT parent virus (3) .

CD is a well-characterized enzyme-prodrug system that functions similar to the herpesvirus TK paradigm (9 , 10) . However, CD is an enzyme found in bacteria and fungi and not mammalian cells that can catalyze the hydrolytic deamination of the relatively nontoxic antifungal agent 5-FC to the commonly used chemotherapeutic 5-FU. 5-FU can be further modified by cellular enzymes into potent pyrimidine antimetabolites, which induce cell death by inhibiting thymidine synthetase and DNA and RNA synthesis (11) . 5-FU has also been reported to diffuse freely through cellular membranes to generate a strong direct bystander effect (12) . Reports using human colorectal xenograft murine models indicated that transfection of 4% of the tumor cells with Escherichia coli CD gene led to survival rates of 60% (13) . However, limited success has been reported in several experiments that used selected breast and pancreatic tumor models because such cells were found to be relatively resistant to 5-FU, possibly attributable to defects in downstream cellular pathways that are responsible for the metabolism of this enzyme (14 , 15) . In support of this hypothesis, expression of bacterial or yeast UPRT in mammalian cells was found to greatly enhance the sensitivity of the cells to 5-FU (15 , 16) . UPRT is an enzyme that catalyzes the conversion of uracil and 5-FU directly into UMP and 5-fluoro-UMP, respectively, by bypassing rate-limiting reactions controlled by the cellular enzymes. Coexpression of CD and UPRT has been reported to increase the sensitivity to 5-FC 10–168 times when compared with CD alone, depending on the experimental model used (15, 16, 17) .

To examine whether the CD/5-FC system could be used with rVSV as a vector delivery system, we generated a novel rVSV suicide gene model expressing the E. coli CD/UPRT fusion gene (rVSV-C:U). We show here that the expression of the foreign gene does not detrimentally affect VSV replication and results in the generation of high levels of CD and UPRT enzyme activity after infection of the cell. Importantly, rVSV-C:U retained selective oncolytic activity and displayed effective bystander effect in vivo and in vitro with concomitant tumor cell killing. Our data indicate that incorporating CD-based strategies into rVSV-mediated gene therapy protocols may improve the oncolytic potential of this tumor treatment approach.

MATERIALS AND METHODS

Chemicals.

[³H]Cytosine (9.5 Ci/mmol), [2-¹⁴C]uracil (54 mCi/mmol), 5-phosphoribosyl-1-pyrophosphate, uracil, cytosine, UMP, 5-FC, low EEO type 1 agarose, and 2.3 mm 5-bromo-4-chloro-3-indolyl phosphate were purchased from Sigma Chemical Co. (St. Louis, MO), and streptavidin-alkaline phosphatase was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). hIFN-α-2a and gancyclovir were obtained from Roche Laboratories, Inc. (Nutley, NJ), and 5-FU was purchased from Pharmacia & Upjohn Co (Kalamazoo, MI). Antimouse IFN-γ antibodies and biotinylated anti-IFN-γ monoclonal antibodies were acquired from PharMingen (San Diego, CA). TLCSs and polyethyleneimine-impregnated TLCSs were acquired from EM Science (Gibbstown, NJ), and Nunc MaxiSorp plates were obtained from Nunc (Naperville, IL).

Cell Culture.

Human multiple myeloma cell line 8226/Dox40 was a gift from Dr. L. H. Boise (University of Miami; Ref. 18 ), and the murine T-cell lymphoma EL4 and the murine B-cell lymphoma A20 were gifts from Dr. J. Rosenblatt (University of Miami; Refs. 19 and 20 ). The TSA mammary adenocarcinoma cell line was a gift from Dr. A. Rakmilevich (University of Wisconsin, Madison, WI; Ref. 2 ). B16(F10) melanoma cells, 293T embryonal kidney cells, and the BHK cell line BHK-21 were purchased from American Type Culture Collection (Manassas, VA). HMVEC, a human microvascular endothelial cell line, was acquired from Clonetics Corp. (San Diego, CA). Cells were grown in suspension or in monolayer in the recommended culture medium with fetal bovine serum.

Construction of Recombinant Virus.

The recovery of the infectious recombinant viruses and the growth, purification, concentration, and titration of the WT and recombinant viruses were performed according to the methods described previously (21 , 22) The C:U gene was PCR amplified from the pORF-codA::upp expression vector (InvivoGen). The primers were 5′-GAACTCGAGCCACTATGGTGTCGAATAACGCTT and 3′-GTCGAGCTAGCGAATTTCGACAAGCTTATTTCGTACC, introducing XhoI and NheI restriction sites, respectively. The C:U cDNA was then excised from the amplification product with XhoI and NheI and cloned into the XhoI/NheI site of pVSV-XN2 (23) using a Zero Blunt TOPO shuttle vector (Invitrogen, Carlsbad, CA). The plasmid pVSV-XN2 contains the entire VSV genome and has unique XhoI and NheI sites flanked by VSV transcription start and stop signals. After recovery of the infectious rVSV in BHK cells, monoclonal virus was obtained by plaque purification and further purification, and concentration was achieved by sucrose centrifugation. Virus stock titers were measured by standard plaque assay.

CD Enzyme Assay Measured Spectrophotometrically.

CD activity was measured directly by using cell lysates or indirectly by measuring 5-FU released in the culture media. BHK cells (5 × 10⁶) were infected with rVSV-GFP at MOI = 5.0, rVSV-C:U at MOI = 0.1, or rVSV-C:U at MOI = 5.0. To measure 5-FU in culture media, 5 mm 5-FC was added to the media 2 h postinfection. Samples of culture media (10 μl) were collected at different time points and quenched with 990 μl of 0.1 m HCl. To measure C:U activity using cell lysates, 6 h postinfection cells were washed twice with PBS, resuspended in PBS at a concentration of 5 × 10⁶ cells/100 μl PBS, and exposed to five cycles of thawing/freezing. Fifty μl of each sample were mixed with 950 μl of 3 mm solution of 5-FC in PBS and incubated at 37°C. At different time points, 50-μl samples were collected and quenched with 950 ml of 0.1 mm HCl (24) .

The concentration of 5-FU and 5-FC for each condition was determined spectrophotometrically using the following formulas: 5-FC (mm) = 0.119(A₂₉₀) − 0.025(A₂₂₅); and 5-FU (mm) = 0.185(A₂₂₅) − 0.049(A₂₉₀). Each experiment was performed independently three times.

CD and UPRT Chromatographic Assays.

CD and UPRT enzymatic activities were measured as described previously (14 , 15 , 25) . 293T cells were transfected with C:U gene using the LipofectAMINE system. Confluent BHK cells were infected with rVSV-C:U at MOI = 0.1 or MOI = 5.0. For CD activity, 6 h postinfection BHK-infected cells, 293T-transfected cells, and 293T and BHK control cells were harvested in a mixture of 100 mm Tris (pH 7.5), 1 mm EDTA, and 1 mm DTT at a concentration of 1 × 10⁶ cells/10 μl lysis buffer and frozen. At the time of enzymatic measurements, cells had gone through five freeze/thaw cycles, and cellular debris was removed by centrifugation (5 min at 14,000 rpm). For UPRT activity, BHK-infected cells (6 h postinfection) or control BHK cells were harvested in lysis buffer [50 mm Tris HCl (pH 7.5), 150 mm NaCl, 5 mm EDTA, 1 mm DTT, and 1% Triton X-100] at a concentration of 1 × 10⁶ cells/10 μl lysis buffer and incubated for 30 min at 4°C. Ten μl of cell lysate were mixed with 10 μl of [³H]cytosine (0.5 μCi) for CD activity or with 10 μl of reaction buffer [100 mm Tris-HCl (pH 7.5), 10 mm MgCl₂, 10 mm 5-phosphoribosyl-1-pyrophosphate, and 1 μl of uracil-5,6-³H (1 μCi)] for UPRT activity and incubated at 37°C for 2 h. Ten μl of each reaction and 10 μl of a marker solution (unlabeled cytosine and uracil or uracil and UMP) were loaded on TLCSs for CD activity or on propidium iodide-TLCSs for UPRT activity and chromatographed in butanol/water (86:14). The bands corresponding to cytosine and uracil or to uracil and UMP were cut out under short-wave UV illumination, and the radioactivity was assayed by liquid scintillation counter. Each experiment was performed independently three times.

Kinetics of Cytolytic VSV and One-Step Growth Curve.

BHK-21 cells were infected with WT-VSV, rVSV-GFP, rVSV-TK, or rVSV-C:U at MOI = 10. After 45 min, infection media were removed; cells were washed five times with PBS and covered with fresh media. At the indicated time points, samples of culture media were collected, and viral titers were measured by standard plaque assay (5) . For analysis of virus replication and cytolysis after infection with low-dose virus, BHK cells were plated at 10⁶ cells/well in 6-well dishes and infected 18 h later with MOI = 0.01 of WT-VSV, rVSV-GFP, rVSV-TK, or rVSV-C:U for 45 min, washed once with PBS, and covered with fresh media. At 12,18, 24, and 28 h, cells were examined by microscopy. At 24 and 48 h, cell viability was assayed by trypan blue exclusion, culture media samples were collected, and viral titers were measured by standard plaque assay.

In Vitro Evaluation of the Effect of 5-FU Bioactivation and Effect of 5-FU and 5-FC on Virus Replication.

TSA or BHK cells were plated at a density of 5 × 10⁵ cells in 6-well dishes and treated (in triplicate) with different concentrations of 5-FU or 5-FC (1, 3, 6, and 10 μm and mm, respectively). After 24 h, cells were infected with HI-VSV or rVSV-GFP at MOI = 0.1 or 10. The percentage of infected cells was evaluated at 4, 12, and 24 h by fluorescent microscopy. Samples of culture media collected at 24 h were used to measure viral titers by standard plaque assay. Cell survival was measured by trypan blue staining at 24 h. A20 cells were infected with HI-VSV, rVSV-GFP, or rVSV-C:U at MOI = 0.1. After 4 h, infected cells were mixed in different proportions with noninfected cells at a total concentration of 5 × 10⁵ cells/well in 12-well dishes, and 1 mm 5-FC was added to the media. Cell survival was evaluated by trypan blue staining after 48 h. HMVEC, A20, EL4, 8226/Dox40, or TSA cells were infected with rVSV-C:U at MOI = 0.1 or mock infected (HI-VSV) in the presence or absence of 1 or 3 mm 5-FC (in the case of A20 cells). Samples of culture media from each condition were collected after 24 h and incubated at 60°C for 10 min to inactivate the virus and to preserve the 5-FU activity (26) . Aliquots of treated culture media were mixed in different ratios with corresponding fresh media and added to cells freshly plated at a density of 2 × 10⁵ cells/well in 6-well plates. Cell survival was assayed by trypan blue exclusion after 96 h. Complete inactivation of VSV by heating at 60°C was checked by standard plaque assay.

Virus/Suicide Gene System-Induced Cell Killing.

HMVEC, TSA, B16(F10), 8226/Dox40, or EL4 cells were treated with rVSV-C:U at MOI = 0.1, in the presence or absence of 3 mm 5-FC. Cells treated with 5-FC alone, with HI-VSV, or with WT-VSV at MOI = 0.1 were used as control. Cell survival was evaluated by trypan blue exclusion after 18 h.

IFN Protection Assay.

HMVEC or 8226/Dox40 cells were cultured in duplicate in 6-well dishes at a density of 5 × 10⁵ cells/well, and 12 h later, hIFN-α (1000 UI/ml) was added to the culture media in half of the dishes. After 24 h, culture media were removed, and hIFN-α-treated cells and untreated cells were infected with WT-VSV, rVSV-GFP, rVSV-TK, or rVSV-C:U at MOI = 0.1. Postinfection samples of culture media (24 h) were collected for viral titers, and cell viability was determined by trypan blue exclusion. A20 or control EL4 cells were infected with rVSV-GFP at MOI = 0.1, 10, or 100 and then cultured in 25-cm flasks at a density of 10⁶ cells/ml. Postinfection cells (4, 8, 24, and 48 h) were resuspended, and samples were analyzed by fluorescent microscopy and flow cytometry for calculation of the percentage of GFP-positive cells. A20 cells were infected with rVSV-C:U at MOI = 0.1 in the presence or absence of 1 mm 5-FC. Control cells were treated with 5-FC alone, HI-VSV, or infected with rVSV-TK in the presence of 10 μg/ml GCV or with WT-VSV at MOI = 0.1. Treated cells were cultured at a density of 5 × 10⁵ cells/ml in 25-cm flasks. At 24, 48, and 72 h postinfection, cells were resuspended, and samples were collected for cell survival evaluation by trypan blue staining.

Sensitivity to 5-FU.

HMVEC or TSA cells were plated in 6-well dishes at a concentration of 2 × 10⁵ cells/well. A20 or EL4 cells were cultured in 25-cm flasks (kept upright) at the same density. Twelve h later, increasing concentrations of 5-FU were added to the culture media. After 96 h, cell survival was assessed by trypan blue counting and spectrophotometric measurements at 490. IC₅₀ level was defined as the concentration of 5-FU that induced 50% cell killing compared with control untreated cells (27) .

Experimental Tumor Models.

All in vivo experiments were performed in accordance with the University of Miami animal care guidelines. BALB/c mice (8-week-old) were purchased from The Jackson Laboratory. A20 (10⁶) or TSA (10⁵) cells (in 100 μl of PBS) were injected in the right flank of each mouse. Once tumors were established (4–5-mm diameter), mice were randomized in groups of five and six mice each. One group of mice received five daily i.p. injections of 30 mg/kg 5-FU. The other groups of mice received two injections, 3 days apart, of 2 × 10⁷ of either HI-VSV, rVSV-GFP, rVSV-TK (two groups), or rVSV-C:U (two groups). Starting 24 h after the first virus injection, one group of mice that received an injection with rVSV-TK received 100 mg/kg GCV, daily, i.p. for 7 days, and one group of mice that received an injection with rVSV-C:U received 500 mg/kg 5-FC, daily, i.p. for 10 consecutive days. Tumors were measured every other day using a caliper, and tumor volumes were estimated using the following formula: volume = length × width²/2. Mice were sacrificed when tumors measured >15 mm in any diameter. Results were reported as (a) mean tumor volume of each group at the time point when the first mouse had to be culled and (b) survival of each group of mice over the period of 45 days of follow-up. For statistical analysis, Student’s t test was used (3) . For rechallenge experiments, at the end of 45 days, surviving mice free of A20 cells, a group of five control mice, and the mice that previously received the injection with A20 cells but did not develop tumors were rechallenged with the same number of 10⁶ A20 cells in the contralateral flank.

For 5-FU measurements, TSA tumors were established by s.c. injection of 10⁵ TSA cells in the right flank of each mouse. When the tumors reached approximately 10 mm in diameter, mice were divided into two groups. One group received 90 mg/kg 5-FU i.p., and animals were sacrificed at different time points after the injection. Blood samples and tumors were collected. The other group of mice received one intratumoral injection of 2 × 10⁷ rVSV-C:U followed 24 h later by 500 mg/kg 5-FC i.p. Animals were sacrificed at different time points after having received the i.p. injection, and blood samples and tumors were collected. The 5-FU extraction procedures were performed as described previously (16) . Plasma was separated by centrifugation from blood collected via the retroorbital sinus in heparinized tubes. Fifty μl of plasma were quenched with 500 μl of ethyl acetate/2-propanol/0.5 m acetic acid (84:15:1) and stored at −20°C. At the time of extraction, the samples were thawed, vortexed, and centrifuged. The organic supernatant was evaporated in a Speed Vacuum Concentrator (Savant) and resuspended in 50 μl of mobile phase, vortexed, sonicated for 10 s, centrifuged to remove particles, and analyzed by HPLC. 5-FU and 5-FC were separated isocratically on a Beckman Coulter Gold HPLC System equipped with 32 Karat software and photo diode array 168 detector. The eluates were monitored at 254 and 290 nm. A Whatman Partisil 5 SAX (4.6 × 250-mm) column with matching guard cartridge was used at a flow rate of 0.5 ml/min. The mobile phase was 10 mm NH₄H₂PO₄ buffer (pH 3.8) containing 7% methanol. 5-FU and 5-FC eluated at 6.5 and 7.1 min, respectively.

Viral Titer Measurement in the Tumors.

TSA tumors were established in BALB/c mice by s.c. infliction of 10⁵ TSA cells in the right flank. When tumors reached 10 mm in diameter, mice were randomized in three groups and treated intratumorally with two administrations, 3 days apart, of 2 × 10⁷ HI-VSV (one group) or rVSV-C:U (two groups). One group of mice that received live virus intratumorally were also treated with 500 mg/kg 5-FC i.p. daily. Mice were sacrificed at days 2, 4, and 6, and the tumors were collected, weighed, and frozen. At the time of viral titer measurements, tumors were thawed and homogenized in 1 ml of PBS, and virus was measured by standard plaque assay.

ELISPOT Assay.

We followed the experimental protocol described previously in detail (28) . Nunc MaxiSorp plates were coated with 100 μl/well of a 0.5 μg/ml solution of antimouse IFN-γ antibodies and incubated overnight at room temperature. Spleens were collected from the three groups of two mice each (treated with HI-VSV, rVSV-C:U, or rVSV-C:U + 5-FC) sacrificed on day 6 of the above-described experiment (see “Viral Titer Measurement in the Tumors”). Spleen cells were suspended in 10% RPMI 1640 and plated in the coated wells in triplicate using four different cell concentrations in 2-fold dilutions (from 10⁶ cells/well of the highest concentration). Spleen cells were cultured in the presence or absence of 10⁵ irradiated TSA cells/well. The plates were incubated for 24 h at 37°C in 5% CO₂. The plates were then washed and covered with biotinylated anti-IFN-γ monoclonal antibodies (100 μl/well) and incubated for 90 min at room temperature. After additional washes, 100 μl/well of 0.2 μg/ml of streptavidin-alkaline phosphatase were added, and the wells were incubated for 60 min at room temperature. After washing again, 100 μl of a 1:4 mixture of 3% melted low EEO type 1 agarose and 2.3 mm 5-bromo-4-chloro-3-indolyl phosphate in AMP buffer was added to each well. The developed spots were counted under the dissecting microscope. The triplicate wells with >50 and <100 spots were used to calculate the average of the two experiments ± SD.

RESULTS

Generation and Characterization of VSV-Expressing CD/UPRT.

To evaluate whether VSV could be successfully generated to carry and express the E. coli CD/UPRT (C:U) fusion gene, cDNA-encoding C:U was cloned into the pVSV-XN2 plasmid (pVSV-XN2-C:U), which carries the full-length antigenomic VSV DNA (21 , 22) . The additional foreign transcription unit (C:U) was inserted between the VSV glycoprotein (G) and the polymerase gene (L) in a position flanked by unique XhoI and NheI restriction sites (Fig. 1A) ⇓ . After transfection with pVSV-XN2-C:U, recombinant virus was recovered in BHK cells 48 h later, and plaque purification of resultant recombinant viruses (rVSV-C:U) was carried out as described previously (29) .

To investigate whether the insertion of the foreign suicide gene affected viral replication and infectivity, we analyzed rVSV-C:U production per cell after infection of BHK cells at a MOI of 10. Accordingly, one-step growth curve studies indicated that rVSV-C:U exhibited similar growth characteristics to WT-VSV or rVSV expressing the GFP (rVSV-GFP) or TK (rVSV-TK) and reached 10⁹ viruses/ml within 10 h (Fig. 1B) ⇓ . Indeed, rVSV-C:U induced almost complete cytolysis of BHK cells (MOI = 0.01) within 48 h, similar to WT VSV (Indiana strain), rVSV expressing TK, or rVSV expressing GFP (Fig. 1C) ⇓ . Because of the rapid viral-induced killing, the addition of 5-FC had very little additional effect on the kinetics of BHK cell death (data not shown). The levels of viral production were also similar for rVSV-C:U and other control VSVs, even after infection of BHK cells with very low doses of virus (Fig. 1D) ⇓ . In conclusion, the C:U fusion gene does not appear to affect VSV replication, and rVSV-C:U exhibits in vitro growth characteristics comparable with WT VSV after infection of BHK cells.

VSV Expresses High Levels of Functional C:U.

Confirmation of expression of the C:U gene by rVSV-C:U was next examined by measuring the enzymatic activity of CD using spectrophotometric assay and chromatographic enzyme assays as described previously (15) . Essentially, for chromatographic identification, BHK cells were infected with rVSV-C:U at MOI = 0.1 or 5.0, and cell lysates were retrieved at various times postinfection. This analysis indicated that labeled cytosine was effectively metabolized to uracil in lysates from rVSV-C:U-infected cells but not rVSV-GFP-infected cells (Fig. 2A) ⇓ . Spectrophotometric analysis confirmed expression and indicated that as much as 70 nm 5-FU/min was produced using cell lysates prepared from 1 million cells infected with rVSV-C:U at a MOI of 0.1 or 5 (Fig. 2B) ⇓ . Likewise, high levels of UPRT activity were demonstrated by similar chromatographic analysis of rVSV-C:U-infected BHK cells (Fig. 2C) ⇓ . Collectively, our data indicate that rVSV-C:U expresses high levels of functional C:U.

Analysis of 5-FU Production and Bystander Effect.

A major strength of any prodrug activation model is the potential to extend the cytotoxic therapeutic effect to untransfected target cells. In the case of C:U/5-FC, an efficient bystander effect has been reported because of the production of 5-FU, which can reach neighboring cells by simple diffusion. To evaluate the efficiency of CD catalytic activity, BHK cells were infected with rVSV-C:U at MOI = 0.1 or 5.0 or with rVSV-GFP as a control, in the presence of 5 mm 5-FC. Samples of the infected culture media were collected at different times postinfection, and 5-FU production was measured spectrophotometrically. This analysis indicated that high levels of 5-FU production were detected in the culture media up to 72 h postinfection using rVSV-C:U but not rVSV-GFP (Fig. 3A) ⇓ . Thus, rVSV-C:U-infected cells produce large amounts of CD that are able to catalyze 5-FC into 5-FU beyond 24 h, after which almost all of the cells undergo virus-induced CPE.

Having confirmed the presence of 5-FU in the culture media, we next analyzed the extent of bystander effect. Cultured tumor cells (TSA, EL4, A20, and 8226/Dox40) or normal cells (HMVEC) were thus infected with rVSV-C:U or mock infected in the presence or absence of 5-FC. After 24 h, supernatants from these cells were recovered and incubated at 60°C to completely inactivate residual virus, a consequence that does not affect 5-FU activity (26) . Heat-treated samples of culture medium from the infected or uninfected cells were mixed at different ratios with new media and added to freshly cultured cells. After 96 h, cell survival evaluation by trypan blue exclusion showed significant killing of tumor lines exposed to culture medium from rVSV-C:U + 5-FC-treated cells (Fig. 3B) ⇓ . Cell death (>50%) was observed even at 1:500 dilution of the media, demonstrating potent bystander activity independent of cell-to-cell contact. In contrast, no cell death was observed in cells treated with tissue-cultured medium from rVSV-C:U-infected cells in the absence of 5-FC or with 5-FC alone (Fig. 3C) ⇓ . HMVEC cells and normal murine BALB/c embryonic fibroblasts were significantly less affected by the rVSV suicide gene system, almost certainly because they were nonviable to infection and did not generate 5-FU (HMVEC IC₅₀ to 5-FU was established at 3–4 μm and was similar to that for the following cancer cells tested: 3–4 μm TSA; 4–5 μm 8226/Dox40; 1 μm A20; and 10 μm EL4). Thus, 5-FU generated by rVSV-C:U exhibits the hallmarks of a functional bystander effect in vitro.

Because 5-FU is a chemotherapeutic that acts by inducing cell death, it was plausible that 5-FU could limit the replication and propagation of rVSV-C:U in vitro and in vivo (26) . To address this concern in vitro, we evaluated the sensitivity of cancer cells pretreated with 5-FC or 5-FU to VSV replication and killing. Accordingly, TSA cells were treated with increasing concentrations of 5-FC and 5-FU for 24 h, followed by infection with rVSV-GFP at MOI = 0.1 or 10. The efficiency of VSV infectivity and oncolysis on cells pre-exposed to 5-FU and 5-FC was evaluated by measuring the percentage of infected cells using fluorescent microscopy and monitoring VSV-GFP replication. This was complemented by measuring virus titers and finally by the degree of cell killing by trypan blue exclusion analysis. Our data demonstrated that infection of TSA cells with high titers of rVSV-GFP (MOI = 10) was not influenced by pretreatment with 10 μm 5-FU or 10 mm 5-FC used in the assay (the cells’ IC₅₀ for 5-FU was determined at 3 μm). The percentage of infected cells was 80–90% by 4 h (Fig. 3D) ⇓ . At lower dose infection (MOI = 0.1), the percentage of infected cells at 12 h was slightly lower after 5-FU pretreatment (Fig. 3D) ⇓ , although by 24 h almost 100% of cells were infected (data not shown). Viral production in TSA cells pretreated with 5-FU was <1 log lower than in control cells, remaining at high levels of 2 × 10⁸. Cell death evaluation by trypan blue exclusion showed almost complete cell death within 24 h, both in control and in 5-FC- and 5-FU-pretreated conditions (data not shown). Taken together, these in vitro assays demonstrate that exposure to relatively high levels of 5-FC and 5-FU did not significantly influence VSV replication and oncolytic activity.

Evaluation of the in Vitro Oncolytic Efficiency of rVSV-C:U/5-FC.

We next evaluated the combined oncolytic efficiency of rVSV-C:U/5-FC on different cancer cell lines or normal cells. Accordingly, rVSV-C:U, rVSV-GFP, HI-VSV, or WT-VSV was used to infect TSA, K562, 8226/Dox40, and EL4 cancer cells or normal HMVEC cells at a MOI of 0.1. Results indicated that all recombinant viruses killed the cancer cell lines (>75% cell death) within 24 h (Fig. 4A) ⇓ . In contrast, rVSV-C:U + 5-FC or VSV-GFP was found to induce a low fraction of HMVEC killing (20–25%) when treated similarly (Fig. 4A) ⇓ . Pretreatment with type I IFN completely prevented virus replication in the normal HMVEC cells but not in the cancer cell lines. For example, rVSV-C:U titers were 2000 times lower in infected HMVEC cells than in 8226/Dox40 cells by 24 h postinfection and decreased to an undetectable level in HMVEC cells, but not 8226/Dox40 cells, when pretreated with IFN-α (Fig. 4, B and C) ⇓ . These data confirm earlier findings that at least one potential mechanism of VSV-mediated oncolysis exploits defects in the IFN pathway present in tumor cells (30) . In addition, these data demonstrate the selective oncolytic efficiency mediated by rVSV-C:U/5-FC.

Because rVSV-C:U infection in the absence of 5-FC clearly exhibits a viral-induced oncolytic effect alone, we next examined the contribution of 5-FU in cancer cell killing. For this experiment, mouse A20 lymphoma cells were used because they were found to exhibit different levels of virus-induced CPE when infected with rVSV-GFP and observed by fluorescent microscopy. For reasons that remain to be clarified, a small population of A20 cells was found to remain uninfected, possibly because some cells differentiated into a subpopulation that did not interact efficiently with the virus (Fig. 4D) ⇓ . However, in general, a high level of viral replication was achieved in the infected population, with viral titers reaching 10⁷ by 24 h postinfection (data not shown). Nevertheless, when 5-FC (1 mm) was added to the culture medium harboring A20 cells infected with rVSV-C:U, almost complete cell death could be demonstrated by trypan blue staining after 48 h (Fig. 4E) ⇓ . In contrast, the dose of 5-FC used (1 mm) did not affect A20 cell growth when used alone. To further distinguish between the contribution of the virus alone and 5-FU bioactivation in A20 cell killing, cells were infected with rVSV-C:U, rVSV-GFP, or HI-VSV, and after 6 h, they were mixed at different ratios with uninfected cells and cultured in the presence of 1 mm 5-FC. Cell viability was measured after 48 h. The difference observed between HI-VSV + 5-FC and rVSV-GFP + 5-FC essentially represents the effects of the virus alone, whereas the difference between rVSV-GFP + 5-FC and rVSV-C:U + 5-FC accounts for the additional oncolytic effect generated by 5-FU (Fig. 4F) ⇓ . Collectively, these data indicate increased killing of A20 cells using rVSV-C:U + 5-FU, emphasizing that bystander effects can contribute toward the efficacy of oncolysis.

In Vivo Activity of the rVSV-C:U/5-FC System.

To evaluate the oncolytic effect of the rVSV-C:U system in vivo, we chose to investigate two aggressive s.c. tumor models, A20 and TSA, that differed slightly in their response to 5-FU. For example, the A20 lymphoma model showed a higher sensitivity to 5-FU (IC₅₀ = 1 μm) compared with the TSA mammary prototype (IC₅₀ = 3–4 μm). Our goals were to evaluate the efficiency of the VSV-directed C:U/5-FC model and to compare it with the previously published rVSV-TK/GCV model. Accordingly, s.c. tumors were grown in the left flank of BALB/c immunocompetent mice. When tumors became palpable, HI-VSV, rVSV-C:U, or rVSV-TK was injected intratumorally (2 × 10⁷ plaque-forming units), followed or not followed by the administration of the respective prodrug, 5-FC or GCV. As an additional control, 5-FU was administered daily for 5 days i.p. Virus treatments were repeated once more after 3 days, and tumor volumes were monitored every other day. For these experiments, two end points were established. First, survival was monitored with the specification that mice were sacrificed when the largest tumor diameter reached 15 mm. Second, evaluation of the mean tumor volumes was determined at a time when the first animal in the experiment was sacrificed.

This study indicated that there was a significant reduction in A20 or TSA tumor growth using either rVSV-TK or rVSV-C:U in the absence of prodrug, compared with HI-VSV (P < 0.001; Fig. 5A ⇓ for A20 and Fig. 6A ⇓ for TSA). However, even more potent tumor inhibition was found in the virus suicide gene/prodrug groups (rVSV-TK/GCV and rVSV-C:U/5-FC; Figs. 5A ⇓ and 6A ⇓ ). Student’s t test analysis indicated that animals harboring A20 or TSA and receiving rVSV-C:U/5-FC treatment faired better than those receiving rVSV-C:U alone (P < 0.05). The same analysis for rVSV-TK/GCV did not reach statistical significance, indicating that generation of 5-FU may be a better prodrug system in these tumor models. Statistical significance was seen between the rVSV-C:U + 5-FC group and 5-FU group [P < 0.001 in the TSA experiments (Figs. 5A ⇓ and 6A) ⇓ and P < 0.05 in the A20 experiment]. There was no difference between the HI-VSV and the HI-VSV + 5-FC groups in either of the experiments (data not shown). Indeed, although many of the tumors appeared of similar size in a number of differently treated groups after 14 days, tumors in all groups except for the rVSV-C:U + 5-FC-treated group grew rapidly. Indeed, animals with A20 tumors survived for the duration of the experiment (45 days) when treated with rVSV-C:U + 5-FC (Fig. 5B) ⇓ . In the control groups HI-VSV or HI-VSV + 5-FC, all animals had to be sacrificed after 20 days. By 90 days, four of six animals in the rVSV-C:U + 5-FC group, two of five animals in the rVSV-TK +GCV group, one of five animals in the rVSV-C:U group, and one of five animals in the 5-FU group remained alive and free of tumors. In the TSA model, the same statistically significant difference in survival was observed between the rVSV suicide gene + prodrug groups and the control groups (P < 0.05; data not shown). The combined intratumoral administration of both rVSV-TK and rVSV-C:U (in conjunction with GCV and 5-FC) did not significantly achieve better results than rVSV-C:U + 5-FC in the A20 model alone (data not shown). Collectively, our data indicate that because of increased bystander effect, rVSV-C:U may be a better oncolytic agent than rVSV-TK.

The Induction of Antitumor Immune Responses by rVSV-C:U.

Previous data have demonstrated that lysis of cells, e.g., through virus replication (CPE), can result in the enhanced generation of CTL activity, arguably by facilitating tumor antigen uptake by professional antigen-presenting cells (31 , 32) . To evaluate this, animals exhibiting no tumor formation after 45 days in the previously presented A20 model were rechallenged with the same number of A20 cells injected in the opposite flank of the animal (six of six animals in the rVSV-C:U + 5-FC group, two of five animals in the rVSV-C:U group, three of five animals in the rVSV-TK +GCV group, one of five animals in the 5-FU group, and zero of five animals in HI-VSV, HI-VSV + 5-FC, and rVSV-TK groups). Significantly, none of the rechallenged animals developed tumors at the site of cancer cell readministration, in contrast with a control group of new animals not treated previously, 4 of 5 of which developed tumors within 15 days postimplantation. These data indicate activation of the immune system against A20 cells, which may have contributed toward the antitumor activity of rVSV-C:U/5-FC.

To further characterize the mechanisms of rVSV-C:U/5-FC action, we measured the activation of T cells in the spleens of mice receiving tumor therapy by performing ELISPOT assay. Mice with s.c. TSA tumors were treated with HI-VSV, rVSV-C:U, or rVSV-C:U + 5-FC. Six days after the initiation of treatment, mice were sacrificed, and spleens were collected. Mononuclear cells were isolated, and IFN-γ release by single cells was measured in the presence or absence of radiated TSA cells as a source of in vitro stimulating antigen. We found that there were 30 and 44 times more T cells activated and secreting IFN-γ in the spleens of the mice that were treated with rVSV-C:U and rVSV-C:U + 5-FC, respectively, than in the spleens of mice bearing TSA tumors that received control HI-VSV (Fig. 6B) ⇓ . The number of secretory cells doubled in both viral treatment conditions when they were exposed to TSA cells as a source of antigen. Interestingly, the number of activated T cells was found to be almost double when rVSV-C:U was combined with generated 5-FU. These data indicate a strong immune component in the antitumor efficiency of rVSV-C:U/5-FC therapy.

We next evaluated, by HPLC, 5-FU levels in the blood after one single maximally tolerated dose of 5-FU or after intratumoral administration of rVSV-C:U in the presence of systemic 5-FC. Blood level measurement in samples collected at different time points revealed that systemic 5-FU administration produced high early peaks (up to 4.4 times higher than the rVSV-C:U + 5-FC by 15 min; Fig. 6C ⇓ ). However, exogenously added levels drop immediately, in vivo, so that after 25 min, 5-FU produced by the viral/prodrug treatment was detected at higher concentrations and maintained therapeutic levels over the next hour, thus indicating longevity of the system. We could not find any difference in the viral titers achieved by the rVSV-C:U virus in the presence or absence of prodrug administration when measured on day 4 (already 2 days through 5-FC treatment) or day 6 (after four administrations of 5-FC), suggesting that 5-FC and 5-FU did not affect VSV propagation in the tumor (Fig. 6D) ⇓ .

Thus, we demonstrate the presence of potent viral, chemotherapeutic, and antitumor immune components after treatment with rVSV-C:U/5-FC, indicating that this system may have promise for development as a therapeutic against malignant disease.

DISCUSSION

VSV has been recently characterized as a selective oncolytic virus showing significant in vitro and in vivo potential as a novel therapy for cancer (30) . Previous studies have demonstrated that VSV is capable of inhibiting the growth of tumors harboring common genetic abnormalities such as p53, myc, or ras aberrations, both after local intratumoral administration and after systemic i.v. treatment (33) . Furthermore, VSV has been shown to eradicate tumors in immunocompetent hosts without any substantial virus-associated CPE occurring in normal tissue (33) . Recombinant adaptations of the virus that contain antitumor genes afford the opportunity to further increase the oncolytic efficiency of VSV, with concomitant specificity and attenuation (3) . In this regard, previous studies have shown that the expression of cytokines such as interleukin-4 and suicide genes such as herpes simplex virus-TK increased the ability of VSV to suppress tumor growth, compared with WT-VSV, in melanoma and mammary cell carcinoma models (3) .

A potential problem concerning viral tumor therapy in immunocompetent hosts remains the response of the immune system to virus infection, which may suppress a required oncolytic effect through inhibition of virus spread (35) . One strategy to help avoid this dilemma may be the addition of a system capable of exerting a strong bystander effect, which may enhance the oncolytic efficacy of the virus therapy by eliminating uninfected tumor cells. In contrast to the TK/GCV system, the CD/UPRT/5-FC system may confer a stronger bystander activity that is independent of intercellular junctions (36, 37, 38) . This may be important in the treatment of tumors with down-regulated intercellular gap junctions (36) . Previous attempts to use viral agents in the therapy of cancer have been limited by lack of selectivity, inefficient replication of the virus in the tumor, or apprehension associated with the transforming potential of the virus itself (2 , 39) . Given this, we therefore decided to evaluate the antitumor/bystander potential of the recently characterized E. coli CD/UPRT fusion suicide gene with the actively replicative, selectively oncolytic rVSV as a viral vector (15 , 17) .

The generation of rVSV-C:U was achieved, and impressively high levels of CD and UPRT activity were apparent. Spectrophotometric measurements of CD activity up to 64 ± 7 nm 5-FU/min were produced by approximately 1 million infected BHK cells. This is in comparison with a recently published value of 10 ± 2 nm 5-FU/min produced by 1 million BNL1 hepatocellular carcinoma cells that were retrovirally transduced with the E. coli CD gene (24) . The CD gene from E. coli also compares favorably with results reported previously using the yeast CD/UPRT fusion gene (15) . In addition, our results indicate that rVSV-C:U + 5-FC generates a very strong bystander killing effect. The amount of 5-FU produced by recombinant viruses in the media essentially eradicated >90% of a number of tumor cell types tested. This observation indicates that 5-FU effectively diffuses in and of cells and, unlike TK/GCV system, does not require cell-to-cell contact for bystander activity (36) .

Other potential issues regarding virus gene therapy include taking into consideration the possibility that the rapid oncolytic effect of the virus itself could limit the amount of chemotherapeutic agent produced in the cell. Conversely, the chemotherapeutic drug could conceivably limit viral replication and subsequent propagation in the tumor (40 , 41) . However, we demonstrated high levels of 5-FU production in cancer cells in vitro and intratumorally in vivo. In vitro we showed that CD continues to transform the provided prodrug up to 72 h postinfection. In addition, our in vitro and in vivo data indicate that the generated 5-FU and the administered 5-FC do not significantly inhibit VSV replication, and the viral titers achieved in the tumors by the rVSV-C:U were similar in the presence or absence of prodrug administration. A probable explanation is that VSV is a RNA virus, and its cytoplasmic replication is less affected by the multiple mechanisms by which 5-FU inhibits DNA and RNA synthesis, altering cell growth and survival. This demonstration makes the rVSV a specifically suitable viral vector for the CD/UPRT/5-FC suicide gene therapy.

We have recently indicated that VSV selectively replicates in tumor cells because of such cells harboring a flawed IFN system. In this regard, IFN pretreatment completely protected the normal cells against rVSV-C:U replication and cytolysis, as demonstrated previously for the VSV oncolytic agents (30) . In contrast, VSV oncolysis of tumor cells was extremely prompt and efficient, masking the contribution of the suicide gene system in all cancer cell lines except the mouse A20 B-cell lymphoma cells. In this particular cancer cell model, rVSV-C:U induced almost complete cytolysis only in the presence of the prodrug 5-FC, demonstrating a synergistic effect of the rVSV and the CD/UPRT/5-FC components in oncolysis of the lymphoma cells. This may reflect that some lymphoma cell types, such as chronic lymphocytic leukemia, exhibit selective resistance to VSV infection, perhaps because of tropism issues. Nevertheless, this finding allowed the opportunity to evaluate the bystander effect of the rVSV-C:U system. Accordingly, rVSV-C:U exhibited greater therapeutic oncolytic activity in the presence of systemically administered 5-FC compared with rVSV-C:U virus alone and with the previously published rVSV-TK/GCV system. Similar results were observed in the TSA mammary tumor model. The contribution of the viral oncolysis to suppression of the tumor can be compared with the measured effect of rVSV-C:U alone, in the context of demonstrating similar viral titers in tumors injected with the virus and exposed or not exposed to systemic 5-FC. The role of the generated 5-FU is revealed by the statistically greater therapeutic efficiency of rVSV-C:U + 5-FC compared with rVSV-C:U alone in both tumor models.

Another potential antitumor action invoked by rVSV-C:U/5-FC is the stimulation of the immune system. Robust immune responses have been documented previously in experiments using CD-based therapies, as demonstrated by the presence of intense inflammatory infiltrates found within the treated tumors (42) , production of cytokines (43 , 44) , eradication of distant tumors (45) , reduction of the bystander effect in nude mice (24) , or after sublethal irradiation (46 , 47) . Although virus replication itself induces strong CTL responses against viral antigens presented on the surface of infected cells, it is plausible that CTL responses may be invoked against tumor antigens by this treatment also (48 , 49) . In our analysis, the generation of CTLs after VSV suicide gene treatment was indicated by the lack of any tumor growth in A20 cancer model rechallenge experiments. In addition, ELISPOT analysis confirmed the presence of IFN-γ-producing T cells specific to the tumor only in animals receiving rVSV-C:U treatment.

Collectively, our data indicate that VSV is able to generate high levels of CD and UPRT. In addition, 5-FU was found to participate in the killing of infected tumor cells and to exhibit bystander effect. Future studies could conceivably enhance VSV-based oncolytic activities by combining suicide genes with immunomodulatory cytokines or with radiotherapy, thus exploiting the radiosensitizing potential of 5-FU (50) .