Δ24-hyCD adenovirus suppresses glioma growth in vivo by combining oncolysis and chemosensitization


Replication-competent adenoviruses could provide an efficient method for delivering therapeutic genes to tumors. The most promising strategies among adenovirus-based oncolytic systems are designed to exploit free E2F-1 activity in cancer cells, which in the absence of pRb activates transcription and regulates the expression of genes involved in differentiation, proliferation, and apoptosis. We previously developed Δ24, an E1A-mutant, conditionally replicative oncolytic adenovirus. Here, we examine the ability of a second-generation Δ24 (Δ24-hyCD) engineered to express a humanized form of the Saccharomyces cerevisiae cytosine deaminase gene (hyCD). Real-time quantitative PCR, Western blotting, thin-layer chromatography, and radioisotope quantitative enzymatic assays confirmed the production of a catalytically active hyCD enzyme in the setting of an oncolytic infection in vitro; other experiments assessing local production of 5-fluorouracil and a concomitant bystander effect showed improved cytotoxicity. The IC50 dose of 5-fluorocytosine (5-FC) required for a complete cytopathic effect by the Δ24-hyCD virus was fivefold lower than with Δ24 alone in U251MG and U87MG malignant glioma (MG) cell lines. Intratumoral treatment of mice bearing intracranial U87MG xenografts with Δ24-hyCD+5-FC significantly improved survival, confirming that Δ24-hyCD with 5-FC is a more efficient anticancer tool than Δ24 alone. Histopathologically, Δ24-hyCD replication was accompanied by progressively augmented oncolysis and drug-induced necrosis. These findings demonstrate that Δ24-hyCD with concomitant systemic 5-FC is a significant improvement over the earlier Δ24 oncolytic tumor-selective strategy for therapy of experimental gliomas.


Gliomas are a highly diverse, chemoresistant, and radioresistant group of brain tumors in humans. Malignant gliomas (MGs), the most common subtype of primary brain tumors, are neurologically devastating and are considered one of the deadliest human cancers. In their most aggressive manifestation, glioblastoma multiforme, median survival of those afflicted ranges from 9 to 12 months and fewer than 3% of patients survive 5 years.1 Few gains have been made in increasing survival time even with the use of modern surgical techniques, chemotherapy regimens, and radiation therapy modalities. Despite much research, the persistent lack of agents that specifically target glioma phenotypes reflects limited prospects for finding successful drug therapy. Adding to the frustration is the fact that glioblastoma multiforme is the second most frequently reported histology of human brain tumors, accounting for 23% of all primary brain tumors in adults.1 Developing alternate methods of treatment, especially those based on exploiting the current understanding of the molecular biology of gliomas, is therefore a clear and urgent necessity.

One such molecular-based treatment uses conditional replication-competent adenoviruses that specifically replicate within tumor cells but are unable to replicate (or replicate much less efficiently) in normal cells. Our laboratory previously tested the antiglioma effect of Δ24, a 24-base-pair (bp) deletion mutant E1A adenovirus, a virus whose replication is limited to cancer cells with an abnormal Rb pathway. Targeting the Rb pathway is advantageous in glioma treatment because most gliomas have abnormalities of the p16/Rb/E2F pathway.2, 3, 4, 5 One of the most promising approaches using oncolytic adenoviruses as vectors is combining oncolysis with drug sensitization by using a gene-dependent enzyme/prodrug therapy (GDEPT). Oncolytic systems have also been explored as vectors for delivering exogenous therapeutic genes to cancer cells. Previous work has shown that replication-competent herpes simplex virus (HSV) can be effectively combined with GDEPT strategies to augment the antitumor effects produced by this virus.6, 7, 8, 9 These authors used a thymidine kinase (TK) approach with their HSV as a delivery vehicle. Additionally, Wildner et al10 used a TK approach with an E1B-deleted adenovirus11 to enhance the killing of colorectal cancer cells. Other strategies have also employed E1B-deleted adenovirus mutants for the delivery of TK as well as cytosine deaminase12, 13, 14, 15 as a method to increase tumor cell cytotoxicity and radiosensitization. Extensive work using cytosine deaminase in replication-incompetent adenoviral vectors was described by Miller et al.16 Since the mechanism of replication and the amount of replication differs greatly between E1A and E1B deletion adenoviral mutants, it is unclear if the expression of the transgene within the E3 region of E1A-mutated viruses would result in increased oncolysis or paradoxically blunt the replication of such viruses. It was this uncertainty that led us to develop and test this construct within the context of human MGs.

The use of yeast cytosine deaminase (yCD)17 to produce a “bystander” effect in GDEPT for gliomas relies on the conversion of the relatively nontoxic prodrug 5-fluorocytosine (5-FC) to the cytotoxic drug 5-fluorouracil (5-FU).18 However, this approach has been severely limited by the inability to deliver efficiently a gene to the tumor target. This limitation led to our current hypothesis that combining yCD-based GDEPT with a replication-competent, oncolytic adenovirus could overcome the problem of inefficient gene delivery because the delivery of the CD gene product would be amplified during the infectious process by lateral infection of initially nontransduced cells.

GDEPT strategies are popular, particularly those that capitalize on the conversion of 5-FC to 5-FU, an antimetabolite that has broad activity against several solid tumors, including gliomas.16 One pitfall of this approach has been that 5-FU has a fairly narrow therapeutic index and systemic doses are typically limited by its hematologic and gastrointestinal toxicities. Local production of 5-FU thus has potential advantages for glioma therapy. In this context, intratumoral conversion of a systemically administered 5-FC prodrug could produce local anticancer effects while avoiding systemic toxicity.

Our present endeavors advance the Δ24 oncolytic strategy by successful construction of a replication-competent vector incorporating the restricted Δ24 replicative phenotype, with the conferred ability to transduce a humanized version of Saccharomyces cerevisiae CD (hyCD). Our results showed that Δ24-hyCD expressed a highly active CD enzyme, inducing an improved cytopathic effect over Δ24 in vitro. Δ24-hyCD/5-FC therapy was tested in the most stringent intracranial xenograft model of human glioma for CD/5-FC GDEPT16 and showed superior efficacy relative to Δ24 alone. Both in vitro and in vivo, Δ24-hyCD+5-FC was more efficient than Δ24. From these findings, it is clear that an E1A-mutant oncolytic adenoviral vector can efficiently transduce exogenous genes such as yCD, resulting in a remarkable improvement over previous oncolytic-based therapies for gliomas. The strength of our findings also serves as a “proof of concept” that GDEPT strategies can be successfully linked with E1A-defective oncolytic viral approaches in the quest for an efficacious therapeutic approach for treating the most biologically aggressive and treatment-refractory gliomas.

Materials and methods

Cell lines and culture conditions

U87MG cells (obtained from the American Type Culture Collection, Manassas, VA, cat. #HTB-14) and U251MG human glioma cell lines (kindly provided by Dr Yung's laboratory) were cultured in Dulbecco's modified Eagle's/F12 medium (1:1, vol:vol) (Media Tech, Herndon, VA) containing 5% fetal bovine serum (DIFCO) and 2 nM glutamine. Cells were grown in culture at 37°C and at 5% CO2 without antibiotics and were passaged fewer than 12 times during the experiments.


Construction of Δ24 has been described elsewhere.2 This construct has a 24-bp deletion in the E1A gene (nt 923–946, both included), a region known to be necessary for Rb protein binding,19 corresponding to the amino acids L122TCHEAGF129.

To construct Δ24-hyCD, a humanized form of the yeast cytosine deaminase gene (hyCD) was inserted into the E3 region of the Δ24 adenovirus. We chose yeast CD because of its superior enzymatic kinetics over the traditional bacterial form.18 The yeast CD gene (fcy1) is derived from S. cerevisiae and its product has an approximate catalytic efficiency that is 280 times higher than from the bacterial form of the enzyme. A series of 24 synthesized, overlapping oligonucleotide primers (Midland Certified Reagent Co, Midland, TX) with pairs that were sequentially elongated by PCR was used to construct Δ24-hyCD. The process was repeated with progressively longer pieces until the full-length gene was obtained. In addition, the 5′ most distal oligonucleotide contained an idealized Kozack consensus sequence, a proximal HindIII and a distal XbaI restriction site for cloning. The nucleotide sequence of the synthesized gene was also changed significantly (102 of 460 coding base pairs) to optimize a human codon rather than a yeast codon preference. The full-length synthesized gene was cloned into pcDNA3.1 (Invitrogen) and clones were isolated and subsequently sequenced. Several clones with the DNA sequence of interest were then transiently and stably transfected into U87MG and U251MG glioma cell lines and assayed for enzyme activity, as described below. Suitable clones expressing enzyme activity were then cloned into the E3 region of pBHG10 (Microbix). pBHG10-hyCD and pXC1-Δ24 were cotransfected into 293 cells to allow homologous recombination as described previously.2

The viruses were propagated in 293 cells and purified by ultracentrifugation in a cesium chloride gradient. All viruses were titered using a plaque method as well as optical density measurements and were maintained at −80°C until use. Single lots of adenovirus Δ24 and adenovirus Δ24-hyCD were used in the experiments. As controls, we used Δ24-hyCD that had been inactivated by UV light and cells that had been mock infected with culture medium.


5-FU was purchased from Sigma Chemical Company (St Louis, MO) and 5-FC from SP Pharmaceuticals (Albuquerque, NM).

Real-time quantitative PCR

U251MG and U87MG cell lines were grown to 95% confluence, harvested with 0.25% trypsin/EDTA, replanted into T25 flasks to a total of 2 × 106 cells, and then incubated overnight. Media were aspirated and 2 ml of adenovirus Δ24-hyCD was added at 0.1, 1, 5, 10, or 100 PFU/cell to duplicate samples from a viral stock of 1 × 1011 PFU/ml, and the flasks were incubated for 1 hour with continuous shaking. The virus was aspirated and cells were washed twice with phosphate-buffered saline (PBS). Fresh complete medium containing 10% FBS was replaced and the cells were incubated at 37°C for 24, 48, 72, or 96 hours. At that time, the harvested tissue cultures were washed twice with PBS. Floating cells were saved by centrifugation, immediately frozen, and stored at −80°C before the mRNA was harvested. Then, cell pellets were lysed with Trizol reagent (Life Technologies) and the RNA was purified according to the manufacturer's recommendations for subsequent amplification by TaqMan quantitative RT-PCR as described previously.16 The following primers and probe were used for the amplification and detection of the hyCD transgene: forward sequence, 5′-IndexTermCAACATGAGGTTCCAGAAGGG-3′; reverse sequence, 5′-IndexTermCAGTTCTCCAGGGTGGAGATCT-3′; and TaqMan probe, 5′-IndexTermTCCGCCACCCTGCACGGC-3′.

The primers were labeled with FAM label at the 5′-end and TAMRA label at the 3′-end for yeast CD mRNA. Control primers and probes were used for the ribosomal RNA housekeeping gene S9, a gene with little expression variability in human gliomas.20 The expression of mRNA for hyCD was quantified and reported relative to a stably expressing hyCD clone of the glioma cell line U251MG. Expression levels were determined with the ABI 7000 sequence detection system (Applied Biosystems, Foster City, CA).

Cytosine deaminase enzymatic assays

Separation of uracil from cytosine or 5-FU from 5-FC was achieved by thin-layer chromatography as modified from Rubery and Newton.21 Briefly, aluminum-backed silica gel sheets were used (silica gel 60-F-254, EM Science, Germany). Each sheet was spotted with a total of 5 μl of a reaction mix or standards at 1-μl successive spots, with drying before additional spotting. The gel sheets were then resolved in a chromatography tank containing a mixture of 80% chloroform and 20% methanol. The solvent front was quite rapid, with the sheets being resolved within 2–3 minutes. The separated cytosine and uracil or 5-FC and 5-FU were then visualized with UV excitation at 254 nm. For quantitative enzyme assays, these resolved spots were cut out, placed in scintillation vials, and counted.

To measure enzyme activity, we used a procedure adapted by Hamstra et al.18 Briefly, U251MG or U87MG glioma cells transfected with 10 PFU/cell of virus were harvested after 24 hours in an assay buffer (100 mM Tris pH 7.8, 1 mM EDTA) and freeze-thawed three times. Protein concentrations were assessed by the Bradford method. For the conversion assay, 5-FC at 100 mM was spiked with 0.5 mM tritiated 5-FC (2 μCi/mM) and diluted at various concentrations to 30-μl reaction volumes, and either 0.3 or 0.75 μg of protein extract was added. The reaction mixtures were allowed to incubate for 1, 5, 10, or 15 minutes at 37°C, after which they were quenched by the addition of 1 M acetic acid and placed on ice. The extent of reaction conversion was based on the fraction of produced 5-FU divided by the total counts of both the 5-FC and 5-FU bands. The percentage converted was used to calculate the production of 5-FU per μg of protein extract per min of reaction time. The apparent Km and apparent Vmax values were based on nonlinear regression analysis using Graph Pad's Prism program (Graph Pad Software, San Jose, CA). All assays were carried out in triplicate.

Western blot analysis

U251MG and U87MG cell lines were prepared in six-well plates and treated with Δ24, Δ24-hyCD, or PBS (mock treatment condition) as described above. The cells were harvested at 24, 48, 72, or 96 hours after treatment. Total cell lysates were prepared by incubating cells with 1 × SDS sample buffer (62.5 mM Tris-HCl pH 6.8, 2% (w/v) SDS, 10% glycerol, 50 mM dithiothreitol), and protein concentration was quantified by using the bicinchoninic acid method (Pierce, Rockford, IL) and read on a Beckman spectrophotometer. Protein samples (20 μg) were boiled at 98°C for 5 minutes, and lysates were separated on a 15% SDS-Tris glycine polyacrylamide gel, subjected to electrophoresis at 95 V for 2 hours, and transferred to a nitrocellulose membrane. The membrane was blocked with 3% nonfat milk, 0.05% Tween-20, 150 mM NaCl, and 50 mM Tris (pH 7.5) and incubated with primary antibody for yeast CD (1:500; Biogenesis Inc., Kingston, NH). The secondary antibody was horseradish peroxidase-conjugated anti-sheep IgG (Pierce, Rockford, IL). The membranes were developed according to Amersham's enhanced chemiluminescence protocol (Amersham Corp., Arlington Heights, IL).

Cell viability assays

U251MG and U87MG cell lines were grown to 95% monolayer confluence. The cells were trypsinized and harvested with 0.25% trypsin/EDTA, plated in six-well tissue culture plates, and allowed to adhere overnight at 37°C in 5% CO2 humidified incubators. 5-FU or 5-FC in serial 0.5-log concentrations in 100 μl aliquots was added directly to the cells to achieve final concentrations as described previously.16 Cells were incubated at 37°C for 5 days and cell viability by cellular respiration was determined using 3-(4,5-methylthiazole-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfonyl-2H-tetrazolium) (Promega, Madison, WI) according to the manufacturer's protocol. The cell survival fraction was measured at each drug concentration as the ratio of absorbance at 490 nm relative to untreated cells. This calculation was normalized for background absorbance of the culture medium alone. The cell survival fraction was plotted against the logarithm of the drug concentration, and IC50 values were calculated using a sigmoidal dose–response curve with variable slope in Graph Pad Prism 3.01.

The crystal violet assay was performed as described previously.2 Briefly, cells were seeded at 105 cells/well in six-well plates, allowed to grow for 20 hours, and then infected with Δ24-hyCD, Δ24, or UV-inactivated Δ24-hyCD at 10 MOI. Either 5-FU (at 0.25 mM) or 5-FC (at 0.5 mM) was added to the cultures at different times after infection (0–6 days). Cell monolayers were washed with PBS and fixed and stained with 0.1% crystal violet in 20% ethanol. Excess dye was removed with several water rinses.

In vitro cytotoxicity was quantified by using the tetrazolium salt 3-(4-5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Sigma) to measure cell viability. For this assay, 104 cells were seeded in 96-well microtiter plates and infected 24 hours later with 0, 2, or 5 PFU/cell of Δ24-hyCD or Δ24. A total of 16 wells were seeded with untreated glioma cells as a viability control, and 16 wells containing only complete medium were used as a control for nonspecific dye reduction. 5-FC (at 0.5 mM) or 5-FU (at 0.25 mM) was added to the cultures 1 day after infection. The medium was removed 7 days after Δ24-hyCD or Δ24 treatment, and 100 μl/well (1 mg/ml) of MTT was added to each well. The plates were incubated for an additional 4 hours and then read on a microplate reader at a test wavelength of 570 nm. Quadruplicate wells were used for each condition.

Animal studies

Cell implantation and adenoviral treatment were performed as described previously.22 Briefly, implants were placed in 6- to 8-week-old female NuNu mice by using screw-guide hardware with coordinates of 1 mm anterior and 1.2 mm lateral to the bregma. The mice were allowed to heal for 7 days, after which U87MG glioma cells were injected at a depth of 4 mm in the region of the putamen at a concentration of 5 × 105 cells in 10 μl of PBS. At 3 days after tumor cell implantation, a single intratumoral injection of 1.5 × 108 PFU of Δ24-hyCD, Δ24, or 10 μl of PBS was injected three times total on days 3, 5, and 7 for each experimental group (Fig 5). At 5 or 15 days after the single intratumoral injection, the animals were given i.p. PBS or 5-FC at 500 mg/kg once daily, Monday through Friday, until the mice displayed signs of neurologic dysfunction (primarily a lack of avoidance behavior or being hunched over in a posterior position) or until they were killed. Mice were killed by CO2 inhalation and the brains were collected for histopathologic examination and immunohistochemical staining. Animal studies were conducted in the veterinary facilities of the MD Anderson Cancer Center in accordance with institutional guidelines.

Figure 5

Treatment with Δ24-hyCD+5-FC is shown to improve survival of nude mice with implanted U87MG intracranial xenografts over Δ24 alone. The data are represented as Kaplan–Meier survival curves as presented by the start of U87MG intracranial implantation (day 0) after intratumoral injection (day 3) with three doses of virus on days 3, 5, and 7. The P-value (determined by logrank test) of Δ24 compared to controls is P<.001 and the significant difference of median survival times of Δ24 compared to Δ24-hyCD is P<.002.

Immunohistochemical analysis of xenograft tumor sections

Animal brains were harvested, fixed in formalin, embedded in paraffin, and sections were prepared after initial baking at 60°C for 30 minutes. The sections were blocked with 0.3% H2O2 and 100% methanol for 30 minutes and rinsed in 10 mM PBS with 0.2% Triton X-100. The rinsed sections were then treated for 20 minutes in 1:50 Triton/PBS. The following antibodies were used: anti-hexon antibody (diluted 1:150; Chemicon, Temecula, CA), anti-yCD (diluted 1:150; Biogenesis Inc., Kingston, NH), and anti-E1A (diluted 1:200; Santa Cruz Biotech, Santa Cruz, CA). Sections were then incubated with secondary antibodies at a 1:50 dilution at room temperature for 1 hour. Staining was performed with acid-fast 3, 3′-amino diaminobenzidine tablets (Sigma). The sections were counterstained with 0.01% methanol green.

Statistical analysis

The anticancer effect in vivo was assessed by plotting survival curves according to the Kaplan–Meier method, and survivals among treatment groups were compared by using the log-rank test in Graph Pad Prism.


The Δ24-hyCD adenovirus

To generate Δ24-hyCD, we modified the Δ24 adenovirus genome, which includes a 24-bp deletion in the Rb-binding region of the E1A gene,2 and inserted an expression mini-cassette in lieu of the deleted E3 region. The expression cassette is driven by the human cytomegalovirus promoter placed immediately proximal to the hyCD sequence (Fig 1a). A bovine growth hormone polyadenylation region is immediately distal to the stop codon of the synthesized CD gene. This construct was confirmed by sequencing the mini-cassette using a series of primers that covered the entire sequence with some overlap (data not shown). The altered nucleotide sequence that confers human codon preference was confirmed by sequence analysis and is shown in Figure 1b.

Figure 1

Generation of Δ24-hyCD adenovirus. (a) The hyCD sequence is depicted as the complete nucleotide sequence of yeast CD as well as the nucleotide substitutions (highlighted in bold) for optimized humanized codon preference. A Kozak sequence is placed immediately before the starting codon (italicized). Proximal HindIII and distal XbaI restriction sites are placed between parentheses. (b) Schematic illustration of Δ24-hyCD showing the 24-bp deletion in the E1A region (nucleotides and corresponding amino-acid residues are shown) and the insertion of the modified cytosine deaminase (hyCD) expression mini-cassette in the deleted E3 region.

Production of cytosine deaminase

To determine if Δ24-hyCD expressed a functional exogenous gene, we initially assessed yeast CD messenger (mRNA) expression by using real-time quantitative PCR. U251MG and U87MG glioma cell lines were exposed to various concentrations of Δ24-hyCD and allowed to incubate for various periods of time (Table 1). The production of mRNA (expressed as fold increase over control) was compared to the amounts of mRNA produced by a U251MG stable clone expressing hyCD (the table displays some negative values which reflects less mRNA production than the stable clone at the early time points and in initial low titer experiments). A dose- and time-dependent trend in increasing mRNA production was evident, indicating that mRNA was increasingly produced with increasing virus as well as with longer incubation times. The production of mRNA was abated at the longest incubation times and at the highest viral titers, probably secondary to improved viral oncolysis. In the 100 PFU/cell treatment condition, the vast majority of cells were floating at 96 hours (data not shown). The U251MG cell line generated higher peak concentrations of mRNA than U87MG cells, likely a result of their more efficient transducibility.16 Production of a 0.04 log increase in CD mRNA over that of controls was seen at only 1 PFU/cell and at 96 hours in the U251MG cells. This amount was approximately 10,000 times the production at 48 hours (1.08 log) and is a dramatic increase over that produced at 24 hours (0.69 log). In contrast, mRNA production in the U87MG cell line required 96 hours or a titer of 100 PFU/cell to reach a 2-log increase of mRNA production over controls. In the U251MG cell line, production of mRNA was maximal at a lower PFU/cell input dose (i.e., 1 MOI) but at a long (96 hours) incubation time (5.04 log) (Table 1). This finding further indicates that the replication competency of Δ24-hyCD directly influences the production of an exogenously expressed gene. As expected, the efficiency of hyCD mRNA production was influenced by the differential infectivity of cell lines and by the differential kinetics of viral production in the different cells.

Table 1 Expression of cytosine deaminase after infection with Δ24-hyCD by real-time PCR of CD mRNA

Consistent with the increase in the mRNA production, levels of CD protein were also augmented in a time- and dose-related manner. Western blot analyses of U87MG glioma cells infected with Δ24-hyCD showed a marked increase in the amount of protein expressed relative to mock-infected U87MG controls or to U87MG cells treated with Δ24 alone (Fig 2a). These analyses also revealed that saturation of the expressed or translated gene product is limited by the lysing of cells as the incubation times increase, in combination with large viral input doses. Together, these findings show that Δ24-hyCD efficiently transduced high levels of the hyCD exogenous gene.

Figure 2

Analyses of the expression and enzymatic activity of hyCD. (a) Western blot analyses of the expression of exogenous hyCD in U251MG cells. Expression was apparent by 24 hours after infection in a dose-dependent manner. As expected, mock (M) or Δ24-treatment at aPFU/cell of 100 did not result in the expression of CD. The expression level of actin is showed as a loading control. (b) Thin-layer chromatography analyses of cytosine deaminase. U251MG cells stably transfected with the hyCD gene were treated with cytosine at the indicated times and assessed for hyCD activity. The migrated uracil spot was visualized with ultraviolet excitation at 260λ. Lanes 5 and 6 showed a dose-dependence positive result. Negative (cytosine) and positive (uracil) controls are shown in lanes 1 and 2, respectively.

Our next objective was to determine if exogenous hyCD protein demonstrated enzymatic activity. In initial experiments, hyCD enzyme activity was qualitatively assessed by thin-layer chromatography. No identifiable conversion of cytosine to uracil was evident in uninfected cells, a finding consistent with the lack of this pyrimidine salvage pathway in human cells. However, enzymatic activity in the Δ24-hyCD-treated U251MG cells rapidly converted cytosine into uracil (Fig 2b). Enzymatic activity was verified by using a tritiated cytosine radioisotope. Quantitative enzymatic determination of the 5-FC-to-5-FU conversion resulted in an apparent Vmax of 8.4 (±1.0) μM/min/mg protein and an apparent Km of 0.63 (±0.04) 1/mM for hyCD-expressing U251MG cells. The enzyme activity in crude extracts of cells infected with 10 PFU/cell Δ24-hyCD and harvested 24 hours later is summarized in Table 2. The percent conversion of 5-FC to 5-FU by U87MG cells (49.7±4.8 at 5 minutes and 90.3±6.1 at 15 minutes) is similar to that previously reported for the non-humanized yCD.23 Taken together, these observations indicate that Δ24-hyCD is an efficient vector that can be used to deliver an enzymatically active form of hyCD to glioma cells in vitro.

Table 2 Percentage conversion of 5-FC to 5-FU by 0.75 mg cell extract previously incubated for 24 hours with 10 PFU of Δ24-hyCD per cell

Increased glioma cell sensitivity to 5-FC when expressing hyCD

To demonstrate the increased sensitivity of glioma cells expressing hyCD, U251MG cells were exposed to 5-FC at increasing concentrations and assayed for viability. Cells expressing hyCD were approximately three orders of magnitude more sensitive than parental cells to 5-FC (IC50, 12.4 versus 3094 μg/ml) (Fig 3). This increased sensitivity of glioma cells expressing hyCD suggests that Δ24-hyCD should be superior to Δ24 as an antiglioma agent.

Figure 3

Dose–response curves of parental and hyCD stable-transfected U251MG cells treated with 5-FC, as assessed by cell viability assay. Note the shift to the left of the IC50 curve for the hyCD stable-transfected U251MG cells in the presence of 5-FC. IC50 values for both cultures are indicated.

Comparison of the antiglioma effect of Δ24-hyCD and Δ24

We next infected human glioma cells with Δ24-hyCD or Δ24, with or without 5-FC or 5-FU, to analyze adenovirus- and/or drug-induced cell death. Cell death was determined by crystal violet staining of viable cells. The effect of 5-FU on cell killing was no different for Δ24 versus Δ24-hyCD in U251MG (Fig 4a) or U87MG glioma cell lines (not shown). However, the addition of 5-FC to Δ24-hyCD-transduced cells improved cell killing, presumably because of the hyCD enzymatic activity provided by Δ24-hyCD. The addition of 5-FC to Δ24-transduced cells did not affect cell killing at these incubation times and viral titers. Next, to compare the induction of cell death after the addition of 5-FC to Δ24 and Δ24-hyCD, we used an MTT assay to assess cell death and found that Δ24-hyCD essentially reduced the amount of virus required for cell killing by approximately five-fold (Fig 4b). These findings suggest that there was an additive or perhaps a “bystander” effect produced. To assess accurately the existence of a bystander effect, we infected U87MG cells with Δ24 or Δ24-hyCD for 1 hour, washed the plates, added fresh media with 5-FC, and collected the conditioned medium 24 hours later. We treated this medium with UV to inactivate any viral activity, and then added it to uninfected U87MG cell cultures. No effect was seen in 5-FC-treated cells with either conditioned medium from Δ24 or from UV-inactivated Δ24. In contrast, cell death was evident in cells treated with conditioned medium from cultures treated with Δ24-hyCD+5-FC with UV inactivation (Fig 4c). Therefore, a diffusible substance in the conditioned medium—presumably 5-FU—was responsible for the bystander effect.

Figure 4

In vitro antiglioma effect of Δ24-hyCD. (a) Crystal violet analyses of the cytopathic effect of Δ24-hyCD or Δ24 in U251MG cells treated with either 5-FU or 5-FC. Each well represents a different time period of 5-FU or 5-FC treatment indicated in days. (b) Quantification of the viability of U251MG cells by MTT assay, treated with Δ24 (left) or Δ24-hyCD (right) and 5-FC at the indicated doses. UVi, UV-inactivated adenoviral treatment, 5 MOI. (c) Demonstration of Δ24-hyCD-mediated bystander effect. U251MG cells were treated with Δ24-hyCD or Δ24 (at a PFU/cell of 10) 24 hours after infection, and conditioned media were collected. In one set of experiments, the conditioned media were inactivated by UV to ensure that no replication-competent virus was carried over. Conditioned media at the indicated volumes were transferred to fresh U251MG cultures and incubated for 48 hours. Viability was assessed by MTT assay.

Antiglioma effect in vivo

To compare the therapeutic efficacy of Δ24 and Δ24-hyCD and 5-FC in vivo, we implanted U87MG xenografts intracranially in athymic nude mice. The U87MG cell line was selected because it produces glioma tumors in nude mice with highly predictable pathologic features and growth dynamics. In addition, we have previously shown16 that among three intracranial xenograft models of human glioma, including D54MG, U251MG, and U87MG, U87MG is the most refractory to a CD/5-FC GDEPT strategy. We used an implantable guide-screw system developed in the MD Anderson Department of Neurosurgery22 to perform precise tumor implantation and intratumoral injection of a therapeutic agent. An experiment was performed similar to that previously reported2 with three injections of 1 × 108 plaque-forming units (PFU). The results are shown in Figure 5. Three independent experiments were performed, with six to 10 animals per group in each experiment. A significant improvement in the median survival (P<.001) over control groups (34 and 38 days, respectively) was seen for both Δ24 (55 days) and Δ24-hyCD+5-FC (74 days). In addition, there was a significant difference (P<.002) in the median survival time between Δ24 and Δ24-hyCD+5-FC in these experimental groups.

The anticancer effect of a single injection of 1 × 108 PFU of Δ24 was compared with a single injection of 1 × 108 PFU Δ24-hyCD, which is designed to minimize the oncolytic effect on survival by this rapidly expanding tumor system. Control animals were concomitantly treated with PBS. Some of the treated animals were given 5-FC to determine if an added benefit would result. In all experiments, the median survival time for control groups (PBS or PBS+5-FC) was consistently between 30 and 36 days. For animals treated with Δ24 or with Δ24+5-FC, the median survival times were 35 and 32 days, respectively. Δ24-hyCD treatment without 5-FC led to a median survival time of 33 days, results which did not differ from the Δ24 or the Δ24+5-FC groups at a 0.05 level of significance. For the Δ24-hyCD+5-FC condition, two 5-FC dose schedules were tested, “early” (5 days after viral injection) or “late” (15 days after viral injection). The combined median survival time of animals treated with Δ24-hyCD+5-FC was significantly longer than the median survival of animals treated with Δ24 alone or with 5-FC as well as the control animals (P<.0001) (Fig 6). The experiment was arbitrarily terminated at day 98 by killing the long-term survivors: four animals (40%) in the early-5-FC treatment group and (10%) from the late-5-FC treatment group. The median survival time was no different in the early- versus late-treated groups.

Figure 6

Treatment with Δ24-hyCD+5-FC significantly improves survival of nude mice implanted with U87MG intracranial xenografts. Data are represented as Kaplan-Meier survival curves from the day of U87MG intracranial implantation (day 0) after intratumoral injection (day 3) with a single dose of experimental viral groups or PBS alone controls. 5-FC was administered on day 5 after treatment (E) or on day 15 after treatment (L). At both time points, the combination of Δ24-hyCD-mediated oncolysis and 5-FC effects resulted in a significant increase in survival compared with vehicle, Δ24 alone, or in combination with 5-FC, or Δ24-hyCD alone. The P value (determined by logrank test) compared Δ24-hyCD with the combination of Δ24-hyCD and 5-FC. Shown are results from a representative experiment. The median survival of the animals receiving D24-hyCD treated early (E=5 days) or late (L=15 days) were not different.

Histopathologic examination of the tumors and brains

After the mice were killed, the brains were removed, fixed in formalin, embedded in paraffin, and sectioned. Microscopic examination of control animal brains revealed noninfiltrative tumors growing in a sphere-like pattern, with a high level of cell proliferation, hypervascularity, and absence of necrotic areas. Lateral displacement of hemispheric structures and collapse of the ipsilateral ventricle indicated that a mass effect was responsible for the animals’ deaths. The brains of animals that had survived for long periods, in contrast, showed complete tumor regression. Tumor sequelae, including dystrophic calcification and microcyst formation, were identified at the tumor implantation site in the right caudate nucleus (data not shown).

Examination of the brains of animals treated with Δ24-hyCD that died before the arbitrary 98-day termination point demonstrated that death resulted from the mass effect of voluminous ellipsoid tumors. High magnification examination revealed a pattern of three distinct, concentric tumor zones: (1) an innermost central core consisting of necrosis and cellular debris (a pathologic finding that was not present in control tumors); (2) a middle zone consisting of large numbers of tumor cells that displayed prominent viral inclusions admixed with apparently intact tumor cells; and (3) an outer zone composed of intact tumor cells with a few scattered cells showing signs of infection (Fig 7a). Immunohistochemical analyses revealed that Δ24-hyCD adenovirus was able to transduce late (hexon) genes (Fig 7b), consistent with an active replication process. Notably, immunohistochemical staining for hyCD showed that Δ24-hyCD efficiently transduced exogenous hyCD protein into human glioma cells in vivo (Fig 7c). Expression of the enzyme was detected in the cytoplasm of the glioma cells and correlated with cells infected with the adenoviral vector, as demonstrated by anti-yeast-CD antibody staining of glioma cells within the field of cells characterized by the presence of viral inclusion bodies. To examine the consistency of CD expression, we killed two Δ24-hyCD-treated animals at 7 and 14 days after tumor implantation. Immunostaining for CD was positive in the infected cells at both time points, indicating that Δ24-hyCD was able to transduce high levels of hyCD for at least 2 weeks after treatment in the U87MG animal model. Collectively, these observations demonstrate that Δ24-hyCD could infect and replicate in vivo and, more importantly, that Δ24-hyCD could efficiently and consistently transduce the hyCD gene. The combination of these effects resulted in significantly extending survival time.

Figure 7

In vivo demonstration of Δ24-hyCD viral replication and cytosine deaminase expression. (a) Examination of the brains of animals treated with Δ24-hyCD revealed a progressive spread of the virus from the injection site. (N) Necrosis. (VI) Viral inclusions (with arrows). (T) Tumor. (b) Immunohistochemical detection of adenoviral hexon protein, indicating replication of the oncolytic virus. (N) Normal brain. (T) Tumor. (c) Immunohistochemical analyses to identify the expression of CD in the treated tumors revealed positive expression in the cytoplasm of cells with associated inclusion bodies.


We report here that a tumor-selective oncolytic adenovirus (Δ24-hyCD) can be used to transduce efficiently high levels of an exogenous hyCD gene, resulting in a potent antiglioma effect in vitro and in vivo. Gene therapy has been hampered by various obstacles, particularly delivery to an adequate number of tumor cells, specific targeting of tumor cells, and “bystander effect” extension of lethality to tumor cells that are a modest distance away from the main tumor mass.24 We designed our viral construct to address these issues, specifically to gain tumor selectivity by targeting the defective p16/Rb/E2F pathway in MGs by using mutant Δ24 with an E1A deletion; to improve exogenous gene delivery to tumor cells by using a replication-competent adenovirus; and to induce a bystander effect by using a GDEPT strategy.

Tumors such as MGs, which are intrinsically resistant to radiation therapy and chemotherapy and recur exclusively in a local manner, are excellent candidates for gene therapy strategies. We demonstrate here that an exogenous gene product (hyCD) can be effectively expressed in target tumor cells with the Δ24-hyCD oncolytic virus. The hyCD gene product was expressed at very high levels and was able to actively convert 5-FC into 5-FU, which exhibits superior activity against glioma cell lines.

Notably, adding 5-FC to cell cultures or administering it to animals treated with Δ24-hyCD did not interfere with the oncolytic potential of the virus. Our initial concerns were that the hyCD gene would not have enough time to be efficiently expressed in an oncolytic setting. We were also concerned that adding 5-FC would “poison” the infected producer cells, thereby blunting production of the oncolytic virus. Our results showed that the oncolytic system provided a sufficient window of opportunity to produce the dramatic expression of an exogenous gene without obviously abrogating the effective replication of the oncolytic virus. However, time-dependent tumor explant assays that measure the viral output for this particular system was not carried out. Therefore, the extent of viral replication and its ability to spread through a tumor and in the presence of 5-FC was not directly assessed. We did, however, measure the amount of virus produced in an in vitro setting comparing Δ24 to Δ24-CD by a TCID50 assay with viral inoculation of 0.1 and 1 MOI. We found that there was no significant difference in the amount of viral production between these two viruses (data not shown). We also found that the optimal production of hyCD mRNA is time and dose dependent with respect to the target cells and correlates with their inherent adenovirus transducibility and expression of adenovirus receptors, notably coxsackie-adenovirus receptor.16 Specifically, U251MG cells can produce approximately 3-log higher amounts of mRNA than U87MG cells. This finding may relate to differences in the infectivity of U87MG and U251MG cells.2 We also found that input titer and incubation time strongly affected hyCD mRNA production in a cell-specific manner. Subsequently, production of the exogenous hyCD gene allowed adequate formation of the toxic metabolite 5-FU, conferring a true bystander effect. These results were confirmed by pathologic examination of the long-term surviving animals. Pathologic sections of brains from animals treated with Δ24-hyCD recapitulate the characteristic process of “zonal” replicative advancement of an infectious “wave” propagating through the tumor mass, as previously demonstrated by Fueyo et al.2 Moreover, the exogenous genetic burden of the hyCD mini-cassette, with or without 5-FC, did not seem to impede this infectious wave within the tumor.

The ability to assess if an antitumor effect is improved by oncolytic viruses is problematic since there is clearly a dose–response phenomenon involving these agents (J Fueyo and F Lang, personal communication). Animal experiments were performed to assess if the timing of administration of 5-FC was critical; the strategy was to minimize the effect of the oncolytic virus itself by titrating down the input dose to only one rather than three injections. As demonstrated in Figure 6, it appears that the virus has difficulty at this particular dose in “catching-up” to the rapidly expanding tumor.

At this particular dose, we are able to better assess the effect of the exposure to converted 5-FC with exposure to 5-FU alone and to characterize the resulting bystander effect. Surprisingly, there was no significant difference in the median survival between administrating 5-FC at day 5 versus 15. Although a late “tail” effect on long-term survival is suggested with some potential benefit for the early treatment group, large animal studies with very long endpoints would be needed to verify this suspicion since the differences appear quite small. These more speculative and longer-term survival studies are out of the scope and defined purpose of this current study.

The realization that gene delivery strategies can be effective when an oncolytic or replication-competent virus is used makes devising improved therapeutic strategies plausible. Our previous work2 showed that an attenuated virus constructed by negatively modulating the interaction between E1A proteins and components of the Rb pathway within the E1A region effectively targets cells with p16/Rb/E2F pathway defects. At least 95% of gliomas demonstrate mutations of this critical pathway; thus, our strategy provides excellent selectivity for treating this devastating disease.25

Since our characterization of the Δ24 system,2 several reports have documented improved infectivity by the insertion of an RGR motif in the fiber knob,26 improved selectivity in an environment in which cells are actively dividing27 and, recently, improvement of the anticancer effect by combining radiotherapy and oncolysis (MLM Lamfers and CMF Dirven, personal communication). The study reported here adds to what has already been learned about the Δ24 system by demonstrating that addition of GDEPT can create a bystander effect. Our findings demonstrate that Δ24 can be used as a delivery vector to produce the combined effect of GDEPT with oncolysis. We also demonstrated that addition of a CD-based therapy to oncolysis improves the anticancer efficacy of Δ24. Of further note is the importance of potential improved safety that the addition of the CD strategy provides to Δ24.


  1. 1

    Central Brain Tumor Registry of the United States. Statistical Report: Primary Brain tumors in the United States, 1992–1997. (2001) Available at http://www.cbtrus.org/reports.htm (accessed November 21, 2002).

  2. 2

    Fueyo J, Gomez-Manzano C, Alemany R, et al. A mutant oncolytic adenovirus targeting the Rb pathway produces anti-glioma effect in vivo. Oncogene. 2000;19:2–12.

    CAS  Article  Google Scholar 

  3. 3

    Alemany R, Gomez-Manzano C, Balague C, et al. Gene therapy for gliomas: molecular targets, adenoviral vectors, and oncolytic adenoviruses. Exp Cell Res. 1999;252:1–12.

    CAS  Article  Google Scholar 

  4. 4

    Henson JW, Schnitker BL, Correa KM, et al. The retinoblastoma gene is involved in malignant progression of astrocytomas. Ann Neurol. 1994;36:714–721.

    CAS  Article  Google Scholar 

  5. 5

    Huang HJ, Yee JK, Shew JY, et al. Suppression of the neoplastic phenotype by replacement of the RB gene in human cancer cells. Science. 1988;242:1563–1566.

    CAS  Article  Google Scholar 

  6. 6

    Aghi M, Chou TC, Suling K, et al. Multimodal cancer treatment mediated by a replicating oncolytic virus that delivers the oxazaphosphorine/rat cytochrome P450 2B1 and ganciclovir/herpes simplex virus thymidine kinase gene therapies. Cancer Res. 1999;59:3861–3865.

    CAS  PubMed  PubMed Central  Google Scholar 

  7. 7

    Chase M, Chung RY, Chiocca EA . An oncolytic viral mutant that delivers the CYP2B1 transgene and augments cyclophosphamide chemotherapy. Nat Biotechnol. 1998;16: 444–448.

    CAS  Article  Google Scholar 

  8. 8

    Boviatsis EJ, Park JS, Sena-Esteves M, et al. Long-term survival of rats harboring brain neoplasms treated with ganciclovir and a herpes simplex virus vector that retains an intact thymidine kinase gene. Cancer Res. 1994;54:5745–5751.

    CAS  Google Scholar 

  9. 9

    Hermiston T . Gene delivery from replication-selective viruses: arming guided missiles in the war against cancer. J Clin Invest. 2000;105:1169–1172.

    CAS  Article  Google Scholar 

  10. 10

    Wildner O, Blaese RM, Morris JC, et al. Therapy of colon cancer with oncolytic adenovirus is enhanced by the addition of herpes simplex virus-thymidine kinase. Cancer Res. 1999;59:410–413.

    CAS  PubMed  PubMed Central  Google Scholar 

  11. 11

    Bischoff JR, Kirn DH, Williams A, et al. An adenovirus mutant that replicates selectively in p53-deficient human tumor cells. Science. 1996;274:373–376.

    CAS  Article  Google Scholar 

  12. 12

    Freytag SO, Stricker H, Pegg J, et al. Phase I study of replication-competent adenovirus-mediated double-suicide gene therapy in combination with conventional-dose three-dimensional conformal radiation therapy for the treatment of newly diagnosed, intermediate to high-risk prostate cancer. Cancer Res. 2003;63:7497–7506.

    CAS  PubMed  PubMed Central  Google Scholar 

  13. 13

    Rogulski K, Wing MS, Paielli DL, et al. Double suicide gene therapy augments the antitumor activity of a replication-competent lytic adenovirus through enhanced cytotoxicity and radiosensitization. Hum Gene Ther. 2000;11:67–76.

    CAS  Article  Google Scholar 

  14. 14

    Bernt KM, Steinwaerder DS, Ni S, et al. Enzyme-activated prodrug therapy enhances tumor-specific replication of adenovirus vectors. Cancer Res. 2002;62:6089–6098.

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15

    Ueda K, Iwahashi M, Nakamori M, et al. Carcinoembryonic antigen-specific suicide gene therapy of cytosine deaminase/5-fluorocytosine enhanced by the Cre/loxP system in the orthotopic gastric carcinoma model. Cancer Res. 2001;61:6158–6161.

    CAS  PubMed  PubMed Central  Google Scholar 

  16. 16

    Miller CR, Williams CR, Buchsbaum DJ, et al. Intratumoral 5-fluorouracil produced by cytosine deaminase/5-fluorocytosine gene therapy is effective for experimental human glioblastomas. Cancer Res. 2002;62:773–780.

    CAS  Google Scholar 

  17. 17

    Erbs P, Exinger F, Jund R . Characterization of the Saccharomyces cerevisiae FCY1 gene encoding cytosine deaminase and its homologue FCA1 of Candida albicans. Curr Genet. 1997;31:1–6.

    CAS  Article  Google Scholar 

  18. 18

    Hamstra DA, Rice DJ, Fahmy S, et al. Enzyme/prodrug therapy for head and neck cancer using a catalytically superior cytosine deaminase. Hum Gene Ther. 1999;10:1993–2003.

    CAS  Article  Google Scholar 

  19. 19

    Whyte P, Williamson NM, Harlow E, et al. Cellular targets for transformation by the adenovirus E1A proteins. Cell. 1989;56:67–75.

    CAS  Article  Google Scholar 

  20. 20

    Blanquicett C, Gillespie GY, Nabors LB, et al. Induction of thymidine phosphorylase in both irradiated and shielded, contralateral human U87MG glioma xenografts: Implications for a dual modality treatment using capecitabine and irradiation. Mol Cancer Ther. 2002;1:1139–1145.

    CAS  PubMed  Google Scholar 

  21. 21

    Rubery ED, Newton AA . A simple paper chromatographic method for separation of methylated adenines and cytosine from the major bases found in nucleic acids. Anal Biochem. 1971;42:149–154.

    CAS  Article  Google Scholar 

  22. 22

    Lal S, Lacroix M, Tofilon P, et al. An implantable guide-screw system for brain tumor studies in small animals. J Neurosurg. 2000;92:326–333.

    CAS  Article  Google Scholar 

  23. 23

    Kievit E, Bershad E, Ng E, et al. Superiority of yeast over bacterial cytosine deaminase for enzyme/prodrug gene therapy in colon cancer xenografts. Cancer Res. 1999;59:1417–1421.

    CAS  Google Scholar 

  24. 24

    Roth JA, Cristiano RJ . Gene therapy for cancer: what have we done and where are we going? J Natl Cancer Inst. 1997;89:21–39.

    CAS  Article  Google Scholar 

  25. 25

    Ueki K, Ono Y, Henson JW, et al. CDKN2/p16 or RB alterations occur in the majority of glioblastomas and are inversely correlated. Cancer Res. 1996;56:150–153.

    CAS  PubMed  Google Scholar 

  26. 26

    Suzuki K, Fueyo J, Krasnykh V, et al. A conditionally replicative adenovirus with enhanced infectivity shows improved oncolytic potency. Clin Cancer Res. 2001;7:120–126.

    CAS  PubMed  PubMed Central  Google Scholar 

  27. 27

    Fueyo J, Alemany R, Gomez-Manzano C, et al. Preclinical characterization of the antiglioma activity of a tropism-enhanced adenovirus targeted to the retinoblastoma pathway. J Natl Cancer Inst. 2003;95:652–660.

    CAS  Article  Google Scholar 

Download references


We wish to thank Joann Aaron (Department of Neuro-Oncology) and Christine Wogan (Department of Scientific Publications) at MD Anderson Cancer Center for editorial services. We also acknowledge the technical assistance of the animal core facility staff at the MD Anderson Brain Tumor Center. This work was supported in part by grants from the Anthony Bullock III Foundation, the Jonsson Family Foundation and the Golfers Against Cancer organization.

Author information



Corresponding author

Correspondence to Charles Conrad.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Conrad, C., Miller, C., Ji, Y. et al. Δ24-hyCD adenovirus suppresses glioma growth in vivo by combining oncolysis and chemosensitization. Cancer Gene Ther 12, 284–294 (2005). https://doi.org/10.1038/sj.cgt.7700750

Download citation


  • oncolytic virus
  • Δ24
  • cytosine deaminase
  • suicide gene

Further reading


Quick links