Use of replication-competent retroviral vectors in an immunocompetent intracranial glioma model

Abstract

Object.

The authors had previously reported on a replication-competent retrovirus (RCR) that has been demonstrated to be stable, capable of effective transduction, and able to prolong survival in an intracranial tumor model in nude mice. The purpose of this study was further investigation of this gene therapy option.

Methods.

The transduction efficiency of RCR in RG2, an immunocompetent intracranial tumor model, was tested in Fischer 344 rats. The immune response to the RCR vector was expressed by the quantification of CD4, CD8, and CD11/b in tumors. The pharmaceutical efficacy of the suicide gene CD in converting prodrug 5-fluorocytosine (5-FC) to 5-fluorouracil (5-FU) was measured using fluorine-19 nuclear magnetic resonance (¹⁹F-NMR) spectroscopy. Animal survival data were plotted on Kaplan–Meier survival curves. Finally, the biodistribution of RCR was determined using quantitative real-time polymerase chain reaction (RT-PCR) for the detection of retroviral env gene.

There was no evidence of viral transduction in normal brain cells. Neither severe inflammation nor immunoreaction occurred after intracranial injection of RCR-green fluorescent protein compared with phosphate-buffered saline (PBS). The ¹⁹F-NMR spectroscopy studies demonstrated that RCR-CD was able to convert 5-FC to 5-FU effectively in vitro. The infection of RG2 brain tumors with RCR-CD and their subsequent treatment with 5-FC significantly prolonged survival compared with that in animals with RG2 transduced tumors treated with PBS. In contrast to the nude mouse model, evidence of virus dissemination to the systemic organs after intracranial injection was not detected using RT-PCR.

Conclusions.

The RCR-mediated suicide gene therapy described in this paper effectively transduced malignant gliomas in an immunocompetent in vivo rodent model, prolonging survival, without evidence of severe intracranial inflammation, and without local transduction of normal brain cells or systemic organs.

Despite recent improvements in surgery, radiation therapy, and chemotherapy, malignant gliomas still cannot be cured. The mean life expectancy in patients with glioblastoma multiforme is still approximately 1 year.^14,15 Gene therapy offers tremendous promise for the future of cancer treatment. This cutting-edge technology takes direct advantage of our new understanding of cancer at the molecular level, and has been exploited to develop new strategies for killing cells selectively or arresting their growth. The first gene therapy clinical trial for malignant gliomas used a replication-incompetent retrovirus carrying the herpes simplex virus–thymidine kinase gene (known as RIR-HSV-TK), which demonstrated limited effectiveness because of the poor intratumoral transduction.¹⁴ The low gene transduction efficiency of current viral vectors remains a significant barrier to the success of cancer gene therapy. Efficient and long-lasting gene delivery is the major challenge in the development of vectors for gene therapy.^20,28,36 Nevertheless, gene therapy for malignant gliomas still holds tremendous promise. Because the majority of the normal cells in the brain of adult humans are in a postmitotic, quiescent, nondividing state, gene transduction can be restricted to the dividing tumor cells themselves, with no detectable spread or integration into normal brain tissue.

We have developed a modified RCR vector based on the Moloney MuLVs, which are derived from RNA viruses and possess the classic properties of retroviruses.^23,36 We inserted an IRES–transgene cassette between the env gene and the 3′ long terminal repeat of the Moloney MuLV clone, and found that the resulting vector replicated with kinetics similar to those of wild-type Moloney MuLV, and was stable through multiple serial passages in cultured cells. Injection of this RCR vector into established subcutaneous and intracranial tumors in mice resulted in highly efficient transmission of the transgene and, in some cases, transduction of the entire tumor. These results demonstrate the potential utility of RCR vectors for brain cancer gene therapy.¹⁹

The most commonly used strategy in cancer gene therapy has been in the field of suicide gene delivery. This approach allows the administration of a nontoxic prodrug, which is then converted to an active toxic metabolite by a specific enzyme introduced into the target cells by a suicide gene. The toxic agent and metabolites can also diffuse into neighboring cells, creating a so-called bystander effect.⁹ Members of different prodrug activators have been evaluated, with the following results: herpes simplex virus–thymidine kinase converts ganciclovir to ganciclovir triphosphate and blocks DNA synthesis;²¹ CD converts 5-FC to 5-FU, a pyrimidine antagonist that blocks both DNA and RNA synthesis;^8,34 cytochrome P450 converts cyclophosphamide to phosphoramide mustard DNA alkylating agent, which blocks DNA synthesis;³⁹ purine nucleoside, a 6-mercaptopurine-DR, converts 6-mercaptopurine to purine antagonist and blocks DNA synthesis and phosphorylation;³⁰ carboxylesterase converts CPT-11 to SN38, a potent topoisomerase inhibitor; and nitroreductase converts nitrobenzyloxycarbonyl to anthracyclines, a DNA cross-linking agent.⁵

To determine the success of these strategies for future clinical trials, these prodrugs must be efficient and selective substrates for the acting enzyme, and must be metabolized to potent toxins to kill the cells that are transfected. The distributive property of the prodrug and its metabolites is important for the bystander effect to achieve maximum cell killing. It is well known that 5-FU is a potent antineoplastic agent in the treatment of colon, rectal, breast, stomach, pancreatic, and lung cancers. The CD/5-FC combination has been proven effective at controlling tumor growth in animals, and is currently being evaluated in several clinical trials.¹¹ More importantly, due to the absence of activity in humans and murine animals, untransfected normal quiescent brain cells do not express Escherichia coli CD, and are immune to 5-FC treatment.

In this study, we used an immunocompetent rodent model to determine first if the RCR vector encoding the GFP gene would elicit an inflammatory response in the immunocompetent host, potentially reducing the effectiveness of repetitive rounds of gene therapy. Second, we used NMR spectroscopy to determine directly the effectiveness with which the RCR-CD vector is able to convert the prodrug 5-FC to 5-FU. Third, RCR biodistribution, intracranially and systemically, was performed in immunocompetent hosts. Last, we investigated whether brain tumor infected with RCR-CD and treated with prodrug 5-FC would achieve the same long-term survival in immunocompetent rats as in our previous immunodeficient mouse xenograft model.

Materials and Methods

Cell Lines and Culture

The human glioma cell line U87 and the rat glioma cell line RG2 were grown in monolayer culture (37°C, 5% CO₂ in air) in Dulbecco Modified Eagle Medium (Gibco-BRL Life Technologies, Inc., Gaithersburg, MD) supplemented with 10% fetal bovine serum and 24 mmol/L NaHCO₃, 30 mmol/L N2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, (pH 7.4), 1 mmol/L sodium pyruvate, and penicillin/streptomycin.

Retroviral Vector Plasmid Construction

Based on the pZAPd-GFP construct, the ecotropic envelope in replication-competent amphotropic Moloney MuLV vector was replaced with the amphotropic envelope from 4070A, was generated by overlap extension PCR, and designated pAZE-GFP.¹⁸ Briefly, the region of the env gene from the unique NsiI site to the termination codon was amplified by PCR and fused to the encephalomyocarditis virus IRES.^10,12,18,27 The plasmid contains an encephalomyocarditis virus IRES-GFP cassette positioned between the env gene and the 3′ long terminal repeat. The long terminal repeat U3 region of pAZE-GFP was replaced with the cytomegalovirus promoter, generating pACE-GFP, which has been described previously. Figure 1 shows a schematic representation of pACE-GFP, the replication-competent provirus portion of the pACE-GFP plasmid, and plasmid pACE-CD, which contains the IRES-CD cassette. The CD gene was amplified from E. coli genomic DNA by using PCR, in which Pfu polymerase (Stratagene, La Jolla, CA) was used to replace the GFP sequence in pACE-GFP, generating plasmid pACE-CD.

Fig. 1.

Schematic drawing showing the structure of RCR vectors, one carrying the marker gene GFP, and the other one carrying its suicide gene CD. The replacement of the U3 region in the 5′ long terminal repeat with cytomegalovirus (CMV) promoter, and the replacement of ectropic env gene with amphotropic env from 4070A are depicted. The transgene GFP or CD cassette was preceded by an IRES, and inserted precisely at the boundary between the env and 3′ untransduced region sequences of this modified Moloney MuLV genome.

Virus Production and Titration

Virus stock was produced by transfection of the human embryonic kidney cell line 293T, U87 human glioma cells, and rat glioma RG2 cells, by using calcium phosphate precipitation with plasmids pACE-GFP and pACE-CD.¹⁸ Virus-containing supernatant was collected 48 hours posttransfection and passed through 0.45-μm syringe filters before use. Viral vectors were titered on the target cells in six-well plates (2 × 10⁵ target cell) in the presence of 4 μg/ml polybrene (Sigma Chemical Co., St. Louis, MO), and 50 μl filtered supernatant was added to the medium at the time of infection. Titer determination was performed on target cells in the presence of azidothymidine (Sigma Chemical Co.) to prevent secondary vector replication.

Seventy-two hours after infection, cells were harvested and suspended in cold PBS. The GFP-positive cells were determined by flow cytometry (Becton Dickinson, Franklin Lakes, NJ), and the viral titer was calculated using the following formula: titer (TU) = positive GFP cells/50 μl × 1000 μl = n × 10³ TU/ml, where TU stands for transducing units.

Comparison of GFP Expression Level in Tumor Cells

The RG2 and U87 cells were infected with ACE-GFP at a multiplicity of infection of 0.05 and were analyzed for GFP fluorescence intensity after both cell populations were fully transduced by ACE-GFP. Histograms were obtained, and show fluorescence intensity compared with cell number (Fig. 2).

Fig. 2.

Histograms showing fluorescence intensity compared with cell number. The expression level of GFP is compared in the tumor cell lines used. The RG2 and U87 cells were infected with ACE-GFP at a multiplicity of infection of 0.05 and then analyzed for GFP fluorescence intensity after both cell populations were fully transduced by ACE-GFP.

Efficiency of Transgene GFP Expression In Vivo

To implant the tumors, 5 × 10³ cells suspended in 5 μl PBS were injected intracranially into 6- to 8-week-old CD Fischer 344 rats. Three days postimplantation, a 10-μl injection of 10⁶ transducing units/ml RCR-GFP was given intratumorally. The rats were killed on the following schedule: 7, 11, 14, and 21 days after the RCR-GFP infection. The brain was removed and washed three times with Hanks balanced salt solution. Three to four pieces of tumor were then randomly selected from around the injection site and quickly minced with scalpels into fragments. This material was then transferred to 50-ml nonvented tissue culture tubes and digested at 37°C by using 1x dispase/collagenase. The tissue was pipetted up and down every 15 minutes to dissociate the tumor cells further, and then passed through a 70-μm nylon mesh filter unit into a 50-ml Falcon tube. Cells were pelleted at 1200 rpm for 8 minutes, washed, and resuspended. The GFP-positive cells were identified using flow cytometry (XL-MCL Flow Cytometer; Beckman Coulter, Fullerton, CA). The percentage of tumor cells transduced by RCR-GFP in intracranial tumor was determined from the mean value in three or four pieces of tumor.

Analysis of 5-FC Metabolism by Using ¹⁹F-NMR Spectroscopy

We used ¹⁹F-NMR spectroscopy (NMR AS 400; Oxford Instruments, Palo Alto, CA) as an analytical technique for the detection of metabolism in fluorine-containing compounds. Briefly, 100% pretransduced human glioma U87 cells and rat glioma cell line RG2 cells were plated and cultured in 100-mm plates. When cells reached 80 to 90% confluency, the old medium was aspirated and replaced with new medium containing 1 mM 5-FC. Cells and medium were harvested separately at different time points: 1, 3, 6, 12, 24, 36, 48, and 60 hours postincubation. The cells were used to determine the concentration of intracellular fluorinated compounds, whereas the culture medium was measured to determine the concentration of extracellular fluorinated species. Because the NMR measurements required 1 or more hours of data acquisition to detect micromolar concentrations, all sample handling and measurements were performed at 4°C to stabilize cells and minimize the breakdown of nucleotides. An equal concentration (1 mM) of pure 5-FC and 5-FU mixture was added to the normal cell culture medium and run as a positive control.^4,7

In Vitro Cell Viability/Cytotoxicity Studies

Cell viability was determined using the MTS assay, for which the soluble tetrazolium salt MTS is used. The MTS is chemically reduced by cells into formazan, which is soluble in tissue culture medium. The measurement of absorbance of formazan is determined by the number of metabolically active cells. Briefly, 5 × 10³ 100% RCR-CD pretransduced rat glioma RG2 cells were plated onto 96-multiwell plates (Costar; Corning, NY). After 24 hours of incubation at 37°C, the old medium was gently aspirated and replaced with new medium containing different concentrations of the prodrug 5-FC, at 1, 25, 50, 100, 250, and 500 μM. The cells were incubated further for 3, 5, and 7 days. Cell viability was analyzed by measuring absorbance of optical density at 490 nm for the formazan produced. All of the analyses were performed in quadruplicate. The plates were also examined under the microscope to assess the degree of cell survival. The surviving fraction of cells was calculated for each assay by the following formula: percent surviving cells = optical density tested samples/optical density control samples × 100.

The control sample reading was obtained from the untreated wells; this was a mean of the readings in eight wells. All treated wells were analyzed at least in quadruplicate, and results were expressed as the mean percentage of surviving cells.

Immunohistochemical Staining

The rat brains were removed and embedded in Tissue-Tek OCT, snap-frozen, and saved at −80°C, and then 7-μM sections were made using a cryostatic machine. The immune and inflammatory response caused by RCR-GFP was evaluated using immunohistochemistry. Briefly, slides were fixed in acetone, quenched in 0.3% H₂O₂, and blocked with 5% nonfat milk for 30 minutes at room temperature. The slides were incubated with 1:200 mouse anti–rat CD4 (W3/25), 1:200 mouse anti–rat CD8 (MRC OX-8), and 1:200 mouse anti–rat CD-11/b (MRC OX-42; all three from Serotec, Ltd., Oxford, United Kingdom) primary antibodies at 4°C overnight, then incubated with 1:200 biotinylated horse anti–mouse immunoglobulin secondary antibody (Vector Laboratories, Burlingame, CA) for 1 hour, and incubated with the standard avidin–biotin complex (Vectastain ABC; Vector Laboratories) at 1:200 for 1 hour at room temperature. All dilutions were performed in cold PBS, and slides were rinsed three times in PBS between steps. Finally, the immunoreactive products were visualized, with (3-amino-9-ethylcarbazole) serving as the chromogen (AEC Kit; Sigma Chemical Co.). Hematoxylin was used for counterstaining, the specimen was mounted with mounting medium, and a coverslip was placed.

Quantification of CD4-, CD8-, and CD11/b-Positive Cells in Stained Slides

Stained slides were quantified²⁴ by analyzing the area of the positive staining for CD4, CD8, and CD11/b. Positive staining was selected under 20-fold magnification with an Olympus Vanos-S light microscope with digital images acquired with an Olympus C-5050 zoom digital camera (5.0-megapixel acquisitions, Olympus America, Melville, NY) and quantified in terms of pixels by using commercially available advanced imaging software (Simple PCI; C-Imaging System, Cranberry Township, PA). To eliminate bias in the study, selection and subsequent analysis of the slides were performed in a blinded fashion. The significance of the data was determined using descriptive statistics (means, standard deviations, 95% confidence intervals) and analysis of variance.

In Vivo Studies

We selected RG2 cells for this study because they are very invasive in growth, and they display almost the same invasion pattern as human gliomas. The RG2 cells are immunologically compatible with their syngeneic host, CD Fischer 344 rats.¹ An animal protocol was approved by the Institutional Animal Care and Use Committee at the University of Southern California, and all procedures were strictly followed. General anesthesia (40 mg/kg ketamine and 8 mg/kg xylazine delivered via intraperitoneal injection) was induced in 15 male 4- to 6-week-old CD Fischer 344 rats (Harlan Breeders, Indianapolis, IN) weighing between 200 and 325 g, and the animals were then placed in a stereotactic head frame (Harvard Apparatus, Holliston, MA). A 0.5-mm bur hole was drilled 2 mm to the right of the sagittal suture and 2.5 to 3 mm posterior to the coronal suture through a scalp incision. Stereotactically guided injections were done using a 10-μl syringe (Hamilton Co., Reno, NV) equipped with a 25-gauge needle and mounted on a Kopf stereotactic apparatus (Kopf Instruments, Tujunga, CA). The needle was inserted into the frontal lobe to a depth of 4 mm, and 5 × 10³ RG2 cells suspended in 5 μl PBS were inoculated. Three days after tumor implantation, a 10-μl injection of 10⁶ IU/ml supernatant freshly harvested from the l00% RCR-CD pretransduced RG2 cells was given stereotactically over 10 minutes. The animals in the control group were injected with 10 μl PBS instead of RCR-CD (RG2 cells pretransduced with RCR-CD were injected as a positive control).

Viruses were allowed to transduce for 10 days before an intraperitoneal injection of 500 mg/kg 5-FC or PBS was given. Animals were killed when they were no longer able to walk spontaneously. The brain and systemic organs, including lung, liver, spleen, kidney, bone marrow, upper GI tract (esophagus), lower GI tract (intestines), skin, and testis were harvested. Brain tumor sizes were obtained after H & E staining. To assess survival, animals were followed until they lost 20% of their body weight and had trouble ambulating, feeding, and grooming, or until other neurological deficits appeared.

Quantitative RT-PCR

For biodistribution and integrated RCR, quantitative RT-PCR targeting the 4070A env gene was performed using the ABI Prism 7700 sequence detector (TaqMan; PE Applied Biosystems, Foster City, CA). The systemic organs from CD Fischer 344 rats bearing rat RG2 glioma tumors infected with RCR-GFP were harvested from different groups at the animals’ planned death and snap-frozen at −80°C. The material, including genomic DNA from brain tumor, lung, liver, spleen, kidney, bone marrow, upper GI tract (esophagus), lower GI tract (intestines), skin, and testis was extracted using the Wizard genomic DNA purification kit (Promega, Madison, WI), according to the manufacturer’s instructions. The probes (PE Applied Biosystems) were chosen with the assistance of the commercially available computer programs Oligo 4.0 (National Biosciences, Inc., Plymouth, MN) and Primer Express 3.0 software (PE Applied Biosystems). For detection of the 4070A env gene, we used forward primer 5′-ACCCTCAACCTCCCCTACAAG-3′, reverse primer 5′-GTTAAGCGCCTGATAGGCTC-3′, and fluorogenic probe 5′-(⁶F-AM)-AGCCACCCCCAGGAACTGGAGATAGA-(TAMRA)-3′.

For precise amounts and qualities of input genomic DNA, we also quantified an internal control gene, ApoB, in each reaction: forward primer 5′-CACGTGGGCTCCAGCATT-3′, reverse primer 5′-TCACCAGTCATTTCTGCCTTTG-3′, and fluorogenic probe 5′-(⁶F-AM)-CCAATGGTCGGGCACTGCTCAA-(TAMRA)-3′. The TaqMan amplification reactions were performed in 50 μl of solution, by using the components supplied in the TaqMan PCR Core Reagent Kit (PE Applied Biosystems). Each sample was analyzed in triplicate, using 250 ng of DNA in each reaction. Thermal cycling was initiated with a 2-minute incubation at 50°C, followed by a first denaturation step of 10 minutes at 95°C, and then 40 cycles of 95°C for 15 seconds and 60°C for 1 minute.²⁹

To quantify copy numbers, plasmid DNA containing the env gene was used to determine a standard curve of plasmid that was assayed for each experiment. Murine fibroblast genomic DNA served as the reference DNA, which was serially diluted and run in parallel to establish the calibration curve and to infer copy numbers from the cycle thresholds, assuming a conversion factor of 6.6 pg of DNA per diploid genome. The normalized gene dose, N, is given by the following ratio: N = copy number of target gene (4070A env)/copy number of reference gene (ApoB) = copy number of env gene/number of cells.

Statistical Analysis

The Student t-test was performed for statistical analysis of in vitro cytotoxicity experiments and intracranial tumor volume. Survival data were analyzed according to the Kaplan–Meier method, using SAS software (SAS Institute, Cary, NC) to calculate significance values. Probability values of 0.05 or less were considered significant.³¹

Results

Immune response and inflammation caused by the viral vectors are known to reduce the effectiveness of gene therapy, even making it useless. The expression of major histocompatibility complex Class-1 in tumor cells is known to increase their immunogenicity and sensitivity to immunomodulators.^22,33 To rule out the possibility that tumor cells themselves could elicit a strong immune response, a rat glioma cell line (RG2) that is syngeneic to Fischer 344 rats was used in our study. It has been demonstrated to have the lowest expression of major histocompatibility complex Class-1 antigen, and is the least immunogenic cell line.²² Quantification of CD8 and CD11/b expression was performed; there was no significant difference between RG2 brain tumors injected with PBS or RCR-GFP, as shown in Fig. 3. No detectable CD4 was found in the tumor tissues.

Fig. 3.

Bar graphs showing quantification of CD8 and CD11/b in intracranial RG2 tumors injected with PBS and RCR-GFP. A: Expression of CD8 in RG2 brain tumors injected with PBS and RCR-GFP. B: Expression of CD11/b in RG2 brain tumors injected with PBS and RCR-GFP.

We have previously demonstrated that RCR vectors can efficiently transduce and stably propagate in malignant human glioma cell lines over multiple infection cycles in vitro. In vivo, RCR was capable of spreading and transmitting an inserted transgene throughout the entire subcutaneous and intracranial U87 glioma in a xenograft mouse model.^32,38 The RG2 cells, on the other hand, had a much slower transduction rate than the U87 cells. Approximately 35% of the tumor may be transduced within 10 days of intratumoral injection; approximately 46% within 2 weeks; and up to 65% after a 3-week infection, as shown in Fig. 4.

Fig. 4.

Biodistribution of the RCR vector carrying marker GFP in RG2 brain tumors implanted in Fischer 344 rats. Approximately 35% of the tumor exhibited GFP expression 11 days after stereotactically guided intratumoral injection of RCR-GFP; approximately 45% expressed GFP by 14 days after viral vector injection; and approximately 65% by 21 days postinjection. Upper Row: Histograms showing the GFP-positive cells. Lower Row: Histograms showing the GFP-positive cells in two dimensions. The RCR-GFP pretransduced RG2 cells were run as a positive control.

To determine if the CD in RCR-CD is functional in the transduced RG2 cells, we performed NMR imaging in these cells and compared it with NMR imaging for transduced U87 cells. Figure 5 demonstrates two resonance peaks in the positive control sample, corresponding to 5-FC (−167.7 ppm), and 5-FU (−169.1 ppm); the fluorine difference is 1.4 ppm. In U87 cells, there is no detectable signal of 5-FU until the 3-hour incubation. After 6 hours, more than half of the prodrug has been converted to 5-FU. By 24 hours, almost all of 5-FC is converted to 5-FU. On the other hand, even after a 60-hour incubation, there is still some 5-FC left in RG2 cells. Interestingly, a broad peak (−164.0 to −164.5 ppm) was detected in the RG2 cells after a 24-hour incubation (data not shown), corresponding to the metabolites of 5-FU (5-fluorouridine and 5-fluoro-2-deoxyuridine).

Fig. 5.

Tracings obtained using the ¹⁹F-NMR modality to detect the pharmacokinetics of CD in converting the prodrug 5-FC to the toxic chemotherapeutic agent 5-FU in 100% RCR-CD pretransduced RG2 and U87 cells in vitro. The fluorine readings shown in the figure were obtained in cell culture medium separated from RG2 (A) and U87 (B) cell cultures after different times of incubation.

The MTS assay demonstrated that the 5-FU produced by RCR-CD-transduced RG2 cells could produce a sharp killing effect, as shown in Fig. 6. Only 10% of cells survived by the 5th day of incubation, and 100% of cells were killed by the cytotoxic agent 5-FU after 7 days of incubation. In vivo, RG2 glioma cells transduced with RCR-CD were treated with 5-FC. The Kaplan–Meier survival curve was plotted using software from Sigma Plot, as shown in Fig. 7. Rats in the control group showed symptoms due to tumor progression at Days 16 to 18, and died within 21 days. Rats infected with RCR-CD and treated with the prodrug 5-FC had a prolonged survival period of up to 35 days (p < 0.001). Rats with 100% RCR-CD pretransduced RG2 cell implants treated with 5-FC showed a significant survival difference (p < 0.001) compared with these two groups. Photomicrographs of tissue stained with H & E showed the tumor pattern in each group. A huge oval tumor (control group), a narrow strip tumor formed inside the striatum (RCR-CD–infected lesions treated with 5-FC), and a very small, rounded tumor (RCR-CD pretransduced and treated with 5-FC) were found.

Fig. 6.

Graph showing results of the MTS assay performed to determine the cytotoxicity of 80% RCR-CD pretransduced RG2 cells treated with prodrug 5-FC at different concentrations (from 0.01–0.5 mM). Approximately 10% of the tumor cells are left after 5 days of incubation; a 100% cell killing effect was achieved after 7 days of incubation with 0.5 mM 5-FC in culture medium.

Fig. 7.

Upper: Graph showing the animal survival rate plotted using a Kaplan–Meier survival curve. The strategy of prodrug administration is shown on the horizontal axis according to the date and the duration of the treatment, labeled as 1^st and 2^nd cycle. All five animals in the negative control group were dead within 21 days. The RCR-CD–infected rats treated with 5-FC can survive significantly longer (up to 35 days) than the animals in the control group. Rats implanted with 100% RCR-CD pretransduced RG2 tumor cells and treated with 5-FC survived significantly longer than the animals in the control group and in the RCR-CD–infected group. Lower: Representative brain sections showing the mean tumor size in each of the three groups. Arrows designate the tumor margins. Original magnification × 1.

To determine that RCR does not have local spread to the contralateral brain and cerebellum, or systemic spread to the bone marrow, spleen, skin, lung, liver, kidney, and lower and upper GI tract, we used RT-PCR to quantitate RCR spread. Because it is possible that hematogeneous spread could result in low-level systemic RCR dissemination, we used RT-PCR to quantitate 4070A env DNA copies in each cell (Table 1). No trace of virus was detected in any of the systemic organs in Fischer rats transduced with RCR-CD intracranially. All the data assessments detailed here were performed in triplicate. The correlation coefficient in each experiment was between 0.99 and 1.

TABLE 1.

Detection of env gene expression by RT-PCR from 4070A in RG2 tumors from brain and systemic organs

	Tumors & Treatment
Organs w/ Tumor	RCR-CD–Infected RG2 & PBS	RCR-CD–Infected RG2 & 5FC	RCR-CD–Pretransduced RG2 & 5FC
brain	+	+	+
lung	−	−	−
liver	−	−	−
spleen	−	−	−
kidney	−	−	−
bone marrow	−	−	−
upper GI (esophagus)	−	−	−
lower GI (intestine)	−	−	−
skin	−	−	−
testes	−	−	−

^*No detectable virus was found outside the central nervous system. The env gene from 4070A was expressed inside the brain tumor only. + = present; − = absent.

Discussion

Our current animal model allows us to evaluate suicide gene therapy strategy according to the following aspects: the vector system for the corresponding gene delivery, the enzyme used for prodrug activation, and the prodrug to be activated. The activation of the prodrug 5-FC by the E. coli CD enzyme has been tested by cell viability/cytotoxicity and NMR spectroscopy. The results demonstrated that CD can function efficiently to convert nontoxic prodrug to potent toxins in in vitro settings, and that it can produce a sharp killing effect in RCR-CD pretransduced cells. Importantly, metabolites of 5-FU also can be detected after administration of prodrug 5-FC.⁶

Although our data showed that only approximately 65% of tumor cells transduced after initial administration of a single viral injection, our results demonstrate that the chemotherapeutic agent 5-FU can easily exit from the transduced cells and enter into the surrounding transduced or nontransduced ones, creating an ideal bystander effect.^16,17

The therapeutic effect in vivo was measured using a Kaplan–Meier survival curve. Animals with RCR-CD infection that are treated with 5-FC can survive significantly longer than control animals (rats treated with PBS instead of 5-FC). Animals with 100% RCR-CD pretransduced RG2 that are treated with 5-FC can survive significantly longer than those in the nonpretransduced RG2 and control groups. Survival is correlated with virus infection in cells. Compared with our previous xenograft U87 nude mouse model, however, the survival time for the immunocompetent rats was much shorter than that of the immunodeficient mice.

There are several possible explanations for this shortened time compared with that reported in our previous publications.^32,38 First, we are using an immunocompetent model. Although we did not detect overt evidence of an immune response in the brain against RCR, the fact that we did not detect RCR systemically in the immunocompetent model suggests that the immune system is capable of interacting with RCR. Recently, Kambara, et al.,¹³ demonstrated that the administration of an immunosuppressant such as cyclophosphamide before on colytic virus administration enhanced the brain tumor infection rate. It is possible that the administration of cyclophosphamide may also enhance the transduction rate of RCR. We are currently exploring this topic. Second, it is possible that different glioma cells have different transduction efficiencies in response to RCR, depending on their retroviral receptor status. The retrovirus uses the phosphate transporter PIT-2 to enter the cell.²⁶ We have not determined whether the PIT-2 receptor varies among different glioma cell lines. Because the Coxsackie adenovirus receptor determines the transduction efficiency of adenovirus, it is possible that the PIT-2 receptor may determine the transduction efficiency of retroviruses.³⁷ Third, in this paper we use a single-shot suicide strategy by activating the suicide gene CD. Nevertheless, chronic treatment with 5-FC may enable these mice to live longer. Fourth, we could use combination suicide genes (that is, TK or CD) to enhance response to therapy. Administration of the prodrugs 5-FC and ganciclovir can be given temporally or simultaneously. Both TK and CD have a bystander effect that can be taken into account.²⁵ Last, RCR therapy could be administered in combination with chemotherapy or radiation treatment. This combination therapy using TK has been performed in the past in in vivo models with good success.^3,15

The safety of retroviral vectors used in clinical trials in humans has been a critical issue since the promise of gene therapy was first recognized. The generation of uncontrolled transduction by RCRs in target cells or tissues is the primary risk associated with the use of retroviral vectors. Systemically, we were able to demonstrate via RT-PCR that there was no evidence of the 4070A env gene from the RCR vector in systemic tissues with highly mitotic cells, such as lung, liver, bone marrow, skin, spleen, kidney, upper and lower GI tract, and testis. We believe that any persistent viruses in immunocompetent hosts may be at least partly controlled by efficient systemic humoral immunity, T cell-mediated cellular immunity, or complement-mediated immunity to the foreign epitopes expressed in target cells.³⁵

In developing future therapeutic vectors, researchers should ensure that these are neither toxic to the target cells nor inducing unwanted side effects, including immunological reactions against the viral vector and its coding sequence.²³ Retroviral vectors may induce an immune response associated with the proliferation of B cells, macrophages (CD11/b), and CD4⁺ and CD8⁺ T cells. The CD8⁺ T cells eliminate unwanted cells by cytotoxic effects and induction of apoptosis or lysis.² After ruling out the possibility of inflammation and immune response from the implantation of tumor cells, no CD4⁺ response was detected in the PBS-and RCR-GFP–injected brain. According to one-way analysis of variance, there was no statistical difference in CD8⁺ and macrophage (CD11/b) expression between intracranial RG2 tumors injected with PBS and those treated with RCR-GFP.

Conclusions

We have demonstrated in vitro and in vivo that RCR-CD is capable of transducing glioma cells in an immunocompetent model. Transduction efficiency was determined by flow cytometry studies for RCR-GFP. Functional efficiency was determined by the in vitro demonstration of conversion of 5-FC to 5-FU. The in vivo effectiveness of RCR-CD was determined using the Kaplan–Meier survival curve. No evidence of systemic spread of RCR was demonstrated with RT-PCR.

Abbreviations used in this paper:

CD

cytosine deaminase

GFP

green fluorescent protein

GI

gastrointestinal

IRES

internal ribosome entry site

MTS

3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium

MuLV

murine leukemia virus

NMR

nuclear magnetic resonance

PBS

phosphate-buffered saline

PCR

polymerase chain reaction

RCR

replication-competent retrovirus

RT-PCR

real-time PCR

5-FC

5-fluorocytosine

5-FU

5-fluorouracil