Yeast Cytosine Deaminase Improves Radiosensitization and Bystander Effect by 5-Fluorocytosine of Human Colorectal Cancer Xenografts¹

In enzyme/prodrug gene therapy, tumor cells are transduced with a gene encoding for a nonmammalian enzyme that converts a nontoxic prodrug into a cytotoxic and/or radiosensitizing drug. The local production of the toxic agent reduces systemic toxicity, which limits the efficacy of conventional chemotherapy.

The two most widely studied enzymes in the enzyme/prodrug strategy are HSV-TK³and bCD. HSV-TK converts GCV into a phosphorylated toxic metabolite (GCV-P) that inhibits DNA polymerase, resulting in DNA chain termination. CD deaminates the prodrug 5-FC into the cytotoxic and radiosensitizing agent 5-FU (1, 2). 5-FU is further metabolized to FdUMP, which inhibits thymidylate synthase resulting in depletion of TTP pools and eventually inhibition of DNA synthesis.

The CD/5-FC strategy has several potential advantages over the HSV-TK/GCV strategy when used for colon cancer treatment. First, the produced 5-FU is the chemotherapeutic agent of choice for treatment of colon cancer. Second, the CD/5-FC strategy appears to have a greater bystander effect. In contrast to GCV-P, 5-FU does not require gap junctional communication to kill nontransduced neighboring tumor cells(3). This is a major advantage, as gap junctions are often lost in cancer cells (4, 5). Third, 5-FU is not only cytotoxic, but also has radiosensitizing properties(6). Indeed, several in vitro and in vivo studies have now demonstrated that the CD/5-FC strategy is more effective than the HSV-TK/GCV system in lung and colon cancer cells (7, 8).

Although the CD/5-FC strategy holds promise, a severe limitation to its effectiveness has come from the fact that 5-FC is poorly deaminated by bCD (9). An enhancement of the conversion efficiency for 5-FC would be expected to substantially improve the efficacy of the CD/5-FC strategy both as a cytotoxic and radiosensitizing treatment. To this end, we have purified and characterized yCD, which has a 22-fold lower K_m for 5-FC compared with bCD. We have demonstrated that the use of yCD significantly improves the efficacy of the CD/5-FC treatment strategy in human colorectal cancer cells in vitro and in vivo (10).

In the present study, we have extended our comparison of bCD and yCD to radiosensitization by 5-FC in CD-transduced human colon cancer HT29 cells in vitro and in vivo. In addition, we have analyzed the bystander effect of yCD- and bCD-transduced HT29 cells, as well as the radiosensitizing effect of 5-FC on bystander cells. Our results demonstrate that yCD produces greater radiosensitization and an increased bystander effect compared with bCD. yCD has, therefore, a high potential to improve the therapeutic outcome of the combined enzyme/prodrug strategy and radiotherapy in cancer patients.

The entire coding sequence of bCD, including an extra Kozak sequence followed by an ATG codon, was isolated by PCR (3). The yCD coding sequence (11), with modifications to facilitate expression in mammalian cells, was kindly provided by Glaxo-Wellcome(10).

The human colon cancer cell line HT29 was grown as described previously(12). Stable HT29 cell lines expressing either bCD or yCD were generated by viral infection (10) using the retroviral expression vector Lazarus (kindly provided by Gary Nolan, Stanford, CA).

The radiosensitizing effect of 5-FC and 5-FU in HT29, HT29/bCD, and HT29/yCD cells was determined using a standard clonogenic assay(13). Cells were treated with 5-FC or 5-FU at various concentrations for 24 h before irradiation at 37°C in media containing 10% dialyzed serum. The radiation survival data were corrected for plating efficiency using a nonirradiated plate treated with 5-FC or 5-FU under the same conditions. The surviving fraction was plotted against the radiation dose, and curves were fit using the linear-quadratic equation. The radiation sensitivity was expressed as the MID, which represents the area under the cell survival curve(14). Radiosensitization was expressed as the ER, which was defined as MID_control/MID_treated.

To determine the cytotoxic and radiosensitizing effect of 5-FC and 5-FU on bystander cells, cocultures of 90% bystander hygromycin-resistant HT29 cells and 10% puromycin-resistant CD-transduced HT29 cells were used. Cell survival of the hygromycin-resistant HT29 cells and puromycin-resistant CD-transduced HT29/cells was determined by plating the cells in selective media after treatment and assessed using a standard clonogenic assay as described above.

In vivo conversion of 5-FC to 5-FU in tumors was monitored by ¹⁹F-MRS. Mice bearing s.c. tumors in the rear limb were injected i.p. with 1000 mg/kg 5-FC and restrained on a specially constructed plastic jig (15) to allow positioning of the tumor under a 6.1-mm single-turn surface coil. Spectra were performed at 282.3 MHz on a Varian Magnetic Resonance System equipped with an 18.3-cm horizontal bore 7.0 Tesla magnet. Spectra were acquired as the average of 277 transients (free induction decays) collected in 8K data points using a 25-μs pulse width(corresponding to a 90° flip angle at a depth of 2.7 mm), a 4.328-ms repetition time, and and a 25-KHz spectral width; 20 min were required for each spectrum. Metabolite concentrations were calculated by normalizing their resonance-peak areas to that of a NaF chemical shift and concentration standard in a microcell placed above the coil. A scaling factor relating the NaF reference signal to ¹⁹F signals arising from the tumor volume was obtained from MRS experiments using the NaF external standard and a tumor phantom containing a known 5-FC concentration.

For the determination of 5-FC and 5-FU levels in plasma, tumors, and normal organs by GC/MS, tumor-bearing mice were injected i.p. with 1000 mg/kg 5-FC. After 90 min, mice were sacrificed, and plasma, tumors,liver, colon and muscle were collected and snap-frozen. Frozen tissue was pulverized and weighed, and homogenates were made in 1 ml of 1 m acetic acid using a polytron. Subsequently, samples were prepared for GC/MS as described previously (16). Quantification of the derivatized products was performed using a Varian Saturn 2000 GC/MS spectrograph in selected ion-monitoring mode. The amounts of 5-FC and 5-FU were expressed as μmol/g tissue.

Nude female mice (Nu/Nu CD-1; Charles River Laboratories,Wilmington, MA) 7–8 weeks of age were injected s.c. in the flank with 5 × 10⁶ viable tumor cells. Tumors were measured with calipers in 2 dimensions. Tumor volumes were calculated using the formula: π/6 (length ×width)²). When tumors measured an average volume of 100–150 mm³, treatment was started. Mice were treated 5 days a week for 2 weeks with either 5-FC (500 mg/kg i.p.), radiation (3 Gy), or a combination of both in which 5-FC was injected 3 h before radiation. The results were plotted as the average tumor volume (in which a cured tumor was included as a tumor of zero size) relative to that at the start of treatment versustime after initiation of the treatment. Differences in the efficacy between treatments were expressed as the number of tumor cures, defined as tumors that did not recur for 60 days, and as the average minimum relative tumor volume for a given condition. Mice were handled according to the established procedures of the University of Michigan Laboratory Animal Maintenance Manual.

Cells were irradiated at room temperature at 1–2 Gy/min using an AECL Theratron 80 (⁶⁰Co). Dosimetry was carried out using a Baldwin ionization chamber connected to an electrometer system that was directly traceable to a National Bureau of Standards calibration (12).

Tumor-bearing mice were restrained in a Lucite restrainer and placed beneath the primary collimator and a secondary heavy alloy collimating device on the (⁶⁰Co) teletherapathy unit. Mice were positioned such that the apex of the tumor was at the center of a 2.4-cm aperture in the secondary collimator (17).

Sections (5 μm thick) of frozen tumor tissues were cut and mounted on slides (Fisher Scientific, Pittsburgh, PA). Tissues were fixed in 4%paraformaldehyde (Electron Microscopy Sciences, Washington, PA) for 10 min at room temperature. Endogenous peroxidase activity was blocked with 0.6% hydrogen peroxide in 100% methanol for 15 min at room temperature. After the slides were washed in water, nonspecific binding sites were blocked with 10% normal goat serum in PBS for 30 min at room temperature. Polyclonal anti-yCD and anti-bCD were generated in rabbits by Berkeley Antibody Company (BabCO, Richmond, CA). Antibodies were diluted 1:200 and 1:75, respectively, in PBS containing 1% BSA and incubated with sections overnight at 4°C. After being washed in PBS (twice for 10 min), sections were incubated with a 1:100 dilution of goat antirabbit IgG (Southern Biotechnology Associates Inc.,Birmingham, AL) for 30 min at room temperature. After 4 further PBS washes, sections were incubated with AEC Histostatin (Zymed, South San Francisco, CA) for 1–5 min at room temperature. The sections were counterstained in Mayer’s hematoxylin for 3 min and mounted in aqueous mounting solution. Randomly chosen regions from at least two tumors were scored for CD-positive cells.

Averages are presented as the mean ± SE. Differences in cytotoxicity and radiosensitivity between cell lines were statistically analyzed with the Student’s t test for unpaired data as were comparisons of minimum tumor size. Complete tumor regressions were compared using the Fisher’s exact test. Statistical significance was defined at the level of P < 0.05 (two-tailed).

We have reported previously that HT29/yCD cells were significantly more efficient at converting 5-FC to 5-FU than HT29/bCD cells, resulting in a greater cytotoxicity (10). We then hypothesized that HT29/yCD cells would be better radiosensitized by 5-FC. To study the effect of 5-FC on the radiation response, we exposed HT29/bCD and HT29/yCD cells for 24 h to varying concentrations of 5-FC,irradiated them, and determined the surviving fraction. As we had hypothesized, the radiosensitizing effect of 5-FC was significantly greater in HT29/yCD cells than in HT29/bCD cells (Fig. 1). At nontoxic concentrations, a ∼3-fold lower concentration of 5-FC was required to radiosensitize HT29/yCD cells to a similar extent as HT29/bCD cells (P < 0.02).

To determine whether the difference in the radiosensitizing effect of 5-FC in HT29/bCD and HT29/yCD cells was attributable to a difference in sensitivity to 5-FU (as opposed to an increased 5-FU production), we also studied the effect of 5-FU on the radiation response of both cell lines. At the nontoxic concentration of 3 μm, both cell lines were radiosensitized by 5-FU, with a slightly higher sensitization of HT29/bCD cells (Fig. 2). At 10 μm, both cell lines were radiosensitized by 5-FU to a similar extent. Additional control experiments showed that HT29/bCD cells were more sensitive than HT29/yCD cells to radiation alone (MID of 2.51 ± 0.19 Gy and 3.74 ± 0.35 Gy, respectively). These results show that the enhanced radiosensitivity produced by 5-FC in HT29/yCD cells was not attributable to a greater sensitivity to radiation or 5-FU.

We then investigated the bystander effect of HT29/bCD and HT29/yCD cells in response to 5-FC treatment. Cocultures consisting of 90%bystander (hygromycin-resistant) HT29 cells and 10% CD-transduced(puromycin-resistant) HT29 cells were incubated with 300 or 1000μ m 5-FC for 24 h, and the surviving fraction of bystander HT29 cells was determined by plating the cells in hygromycin-containing media (Fig. 3 A). For HT29/bCD cells, a small bystander effect was observed when cocultures were treated with 300 μm 5-FC. Increasing the concentration of 5-FC to 1000 μmdid not significantly decrease the surviving fraction of bystander cells. In contrast, a concentration-dependent bystander effect was observed for HT29/yCD cells when cells were exposed to 5-FC, which was significantly greater than that observed for HT29/bCD cells(P < 0.001).

We next studied the radiosensitizing effect of 5-FC on bystander cells(Fig. 3 B). Cocultures of 90% bystander and 10%CD-transduced HT29 cells were treated with 300 or 1000μ m 5-FC for 24 h and subsequently irradiated. No significant radiosensitization was produced in bystander cells after exposure to 300 μm 5-FC when cells were cocultured with HT29/bCD cells, whereas exposure to 1000μ m caused an ER of 1.6 ± 0.2 in bystander cells. When bystander cells were exposed to 300μ m 5-FC in the presence of HT29/yCD cells,significant radiosensitization was produced in bystander cells when compared with HT29/bCD cells (P < 0.01). Increasing the concentration of 5-FC to 1000 μmfurther increased the radiosensitivity of bystander cells.

As the in vitro data provided strong evidence for the superior radiosensitizing effect of 5-FC in HT29/yCD cells, we decided to compare the effect of 5-FC on the radiation sensitivity of HT29/bCD and HT29/yCD tumors grown in nude mice. To obtain a maximum radiosensitizing effect by 5-FC, we first determined the in vivo time course of the radiosensitizing metabolite FdUMP after 5-FC injection. Mice bearing HT29/yCD tumors were injected i.p. with 1000 mg/kg 5-FC and the levels of 5-FC, 5-FU, and metabolites were monitored by ¹⁹F-MRS (Fig. 4 A). Peak levels for 5-FC and 5-FU were observed 60 and 90 min after injection, respectively, and rapidly declined thereafter with half lives of 90 ± 9 (5-FC) and 75 ± 6(5-FU) min. FNuc, of which FdUMP is the major component, were first observed 60 min after injection, reached a steady state approximately 160 min after injection, and remained at that level for at least an additional 4 h.

In addition to HT29/yCD tumors, we also examined the levels of 5-FC,5-FU, and the metabolites in HT29/bCD and HT29 tumors by ¹⁹F-MRS at 100 min after injection of 1000 mg/kg 5-FC (Fig. 4 B). In contrast to HT29/yCD tumors, in which all compounds could be detected, only 5-FC was measurable in HT29/bCD and HT29 control tumors.

To determine whether the absence of the 5-FU peak in HT29/bCD tumors was attributable to a limitation of the detection sensitivity of ¹⁹F-MRS, we also measured 5-FC and 5-FU levels in tumor tissue by GC/MS analysis (Fig. 4 C). Tumor-bearing mice were injected with 1000 mg/kg 5-FC, and 90 min after injection, mice were sacrificed. Plasma, tumors, liver, colon, and muscle were prepared for analysis by GC/MS. For both HT29/bCD and HT29/yCD tumors, 5-FU could be detected, but the concentration of 5-FU in HT29/yCD tumors was approximately 15-fold higher than that in HT29/bCD tumors. The measured amount of 5-FU in HT29/yCD tumors was in correspondence with that observed by ¹⁹F-MRS. In control HT29 tumors, only 5-FC was present.

To study the radiosensitizing effect of 5-FC in vivo, mice bearing HT29/bCD or HT29/yCD tumors were treated with either 5-FC,irradiation, or a combination of both treatment modalities. On the basis of the kinetic data obtained by ¹⁹F-MRS,tumors were irradiated 3 h after 5-FC administration at the time of maximum FdUMP levels in the tumors (see above).

In HT29/bCD tumors, 5-FC treatment caused a small growth delay, and no regressions were observed (Fig. 5). Tumors treated with either radiation or a combination of 5-FC and radiation showed regression, but most regrew after day 30. Tumor cures were observed in 3 of 9 irradiated tumors and in 1 of 10 tumors treated with a combination of 5-FC and radiation. The addition of 5-FC to radiotherapy did not appear to result in a better therapeutic effect in HT29/bCD tumors.

Treatment with 5-FC produced significantly better results in HT29/yCD tumors compared with HT29/bCD tumors (P < 0.005). Tumor regressions were observed, and 8 of 16 tumors were cured(Fig. 5). When HT29/yCD tumors were treated with a combination of 5-FC and radiation, the therapeutic effect significantly improved compared with radiation alone (P < 0.005), with seven of nine tumors being cured. We also assessed the effect of 5-FC and radiation on the relative tumor size. We found that the minimum relative tumor size (compared with pretreatment values) of HT29/bCD tumors treated with 5-FC and radiation was 0.55 ± 0.1,compared with 0.01 ± 0.01 in HT29/yCD tumors(P = 0.002). This improvement was not attributable to a greater sensitivity of HT29/yCD tumors to radiation alone, because their radiosensitivity was similar to that of HT29/bCD tumors (minimum relative tumor volume 0.61 ± 0.21 and 0.65 ± 0.19). These findings confirmed our hypothesis that yCD would produce a greater antitumor effect than bCD when combined with radiation.

As current gene-delivery systems do not transduce all tumor cells, the therapeutic outcome of the CD/5-FC treatment strategy in combination with radiotherapy is largely dependent on the radiosensitizing effect of 5-FU on bystander cells. To study the effect of 5-FC on the radiation response of bystander cells in vivo, we treated mice with tumors consisting of a mixture of 90% bystander and 10%CD-transduced HT29 cells with 5-FC, radiation, or a combination of both modalities, as described above.

Mixed tumors containing HT29/bCD cells did not respond to 5-FC treatment alone (Fig. 6). In addition, consistent with tumors derived entirely from HT29/bCD cells, mixed tumors did not show an improved therapeutic outcome when 5-FC was combined with radiation. Tumor regression was observed with a regrowth after day 35, but no tumor cures were observed. In tumors containing HT29/yCD cells, treatment with 5-FC caused a tumor growth delay, but no regressions were observed. However, the addition of 5-FC to radiation resulted in significantly more tumor regressions when compared with radiation alone, and 6 of 11 tumors were cured(P < 0.025). Similarly, in mice bearing HT29 tumors containing 10% HT29/yCD cells, the combined treatment resulted in a minimum relative tumor size of 0.20 ± 0.07 compared with 0.60 ± 0.1 in 10% HT29/bCD cells(P < 0.001). These results show that 5-FC had neither cytotoxic nor radiosensitizing effects on bystander tumor cells in the presence of HT29/bCD cells, whereas a small cytotoxic effect and a significant radiosensitizing effect of 5-FC were observed in bystander cells in the presence of HT29/yCD cells.

A possible explanation for the greater bystander effect observed in tumors containing yCD cells compared with bCD cells was that yCD cells grew more rapidly in the mixed tumor than bCD cells. To test this possibility, animals were prepared bearing tumors that were 0%, 10%,or 100% HT29/yCD cells or HT29/bCD cells using techniques identical to those described for the therapy experiments. We then assessed the fraction of cells that were CD-positive using immunohistochemistry for bCD and for yCD (Fig. 7). We found that the fraction of positively staining cells in the 10%condition was equal for both the yCD and bCD mixed tumors (although the actual fraction that stained positive was 14%, suggesting that both HT29/yCD cells and HT29/bCD cells may grow faster than bystander cells;Table 1). Thus, the different responses between mixed tumors of bystander cells with either HT29/yCD or HT29/bCD cells is not attributable to a difference in the percentage of CD-expressing cells at the time of 5-FC with or without radiation.

Our results show that, compared with bCD, yCD expression significantly improves the efficacy of combined 5-FC and radiation therapy in human colon cancer xenografts. This improvement can be explained by the higher conversion efficiency of 5-FC to 5-FU by yCD. Moreover, the use of yCD also resulted in greater cytotoxicity to and radiosensitization of bystander tumor cells. As only 5–10% of the tumor cells are transduced with current gene-delivery systems, our findings indicate that yCD is better suited than bCD for cancer gene therapy. This conclusion is strengthened by the fact that the dose-limiting toxicity for 5-FC in humans is attributable to 5-FU production by Escherichia coli in the gut, and that the K_m of yCD for 5-FC is 22-fold lower than that of E. coli CD.

In a previous study we reported on the preferential killing of bCD-expressing tumor cells in response to 5-FC treatment compared with bystander cells, because of the high intracellular 5-FU concentration(3). It was possible that the higher conversion efficiency of HT29/yCD cells for 5-FC, resulting in a higher intracellular concentration of 5-FU, could have killed the “factory” prematurely,thereby reducing the bystander effect of HT29/yCD cells. Our in vitro data showed, however, an increased bystander effect of HT29/yCD cells when compared with HT29/bCD cells, suggesting that the rapid and substantial 5-FU production was sufficient to kill both yCD-transduced producer cells and bystander cells. Nevertheless,temporarily sparing CD-transduced tumor cells from cytotoxicity by intracellular 5-FU might result in a further increase in bystander effect as more 5-FU can ultimately be produced. To increase temporarily the life span of transduced cells, previously we generated a secreted form of bCD.⁴We found that tumor cells that expressed secreted bCD survived longer and produced a higher extracellular concentration of 5-FU in response to 5-FC treatment than cells that expressed intracellular bCD. Thus, we would anticipate that the use of a secreted form of yCD, which is currently being generated, might be even more effective than the intracellular enzyme.

Studies have reported on the radiosensitizing effect of 5-FC in tumor cells transduced with bacterial CD in vitro(18) as well as in vivo (19). A maximal radiosensitizing effect of 5-FC in nude mice bearing WiDr human colon cancer xenografts transduced with bCD was observed after treatment with 5-FC for 6 consecutive days followed by a single radiation dose (19). This is in contrast to our results with HT29/bCD tumors, as we did not observe a radiosensitizing effect of 5-FC. The improved efficacy of the combined treatment in WiDr/bCD tumors might be explained by a possible higher expression of bCD in WiDr cells resulting in a greater sensitivity to 5-FC. Indeed,treatment of WiDr/bCD tumors with 5-FC alone did result in a tumor growth delay, whereas we observed almost no growth delay in HT29/bCD tumors in response to 5-FC treatment. Gabel et al.(19) also showed that the combined treatment of 5-FC and radiation was more effective in tumor-growth inhibition than a similar treatment of the conventional combined treatment of 5-FU and radiation. By using yCD instead of bCD, this difference in efficacy is expected to be higher, which may validate the preferential use of the CD/5-FC strategy over 5-FU when combined with radiotherapy in cancer patients.

Another approach to increase the radiosensitivity of tumors is to combine radiotherapy with double suicide gene therapy, in which tumor cells are transduced with a fusion protein of CD and HSV-1 TK(20). This combined treatment potentiated the antitumor effect when compared with that of the single treatment modalities alone. However, the efficacy of this approach in tumors containing only a small percentage of transduced cells (as will be the case in patients) remains to be determined, because the radiosensitizing and cytotoxic effect of the HSV-TK/GCV on bystander tumor cells is dependent on gap junctional communication, which is often eliminated in tumors.

The first in vivo studies of radiosensitization by 5-FC in adenovirally infected tumors have been published recently (21, 22). In mice bearing human squamous cell carcinoma xenografts(SQ-20B), adenoviral-directed bacterial CD/5-FC gene therapy did enhance the efficacy of fractionated radiotherapy in small and large tumors (21). No tumor cures were observed in large tumors in response to the combined treatment, whereas three of seven small tumors were cured. Because this tumor type was resistant to 5-FU, the improved efficacy was considered to be attributable to an enhanced radiosensitization. In another study (22), human cholangiocarcinoma xenografts (SK-ChA-1) infected with an adenovirus encoding for bCD also showed an improved response to the combined treatment of 5-FC and radiation. On the basis of the current study, the use of yCD would be expected to improve the therapeutic outcome of this treatment strategy in virally infected tumors.

One limitation of our study is that we used athymic nude mice bearing human tumor xenografts. Although this immunocompromised model is required to grow human tumors, it is possible that the immune system could ultimately play a role in tumor response, particularly because CD is not a mammalian enzyme. Additional experiments in an immunocompetent host will be required to address this issue in the future.

In conclusion, our results show an improved therapeutic effect of the combined treatment with 5-FC and radiation when tumor cells are transduced with yCD instead of bCD. This suggests that the combination of radiotherapy and the yCD/5-FC strategy has a high potential to increase the efficacy of radiation in cancer patients. Because the results in mixed tumors were promising, we are currently optimizing adenovirus-mediated delivery of yCD to tumors in a preclinical model of colorectal cancer.