Hypoxia-targeted triple suicide gene therapy radiosensitizes human colorectal cancer cells

The hypoxic microenvironment, an important feature of human solid tumors but absent in normal tissue, may provide an opportunity for cancer-specific gene therapy. The purpose of the present study was to investigate whether hypoxia-driven triple suicide gene TK/CD/UPRT expression enhances cytotoxicity to ganciclovir (GCV) and 5-fluorocytosine (5-FC), and sensitizes human colorectal cancer to radiation in vitro and in vivo. Stable transfectant of human colorectal HCT8 cells was established which expressed hypoxia-inducible vectors (HRE-TK/eGFP and HRE-CD/UPRT/mDsRed). Hypoxia-induced expression/function of TK, CD and UPRT was verified by western blot analysis, flow cytometry, fluorescent microscopy and cytotoxicity assay of GCV and 5-FC. Significant radiosensitization effects were detected after 5-FC and GCV treatments under hypoxic conditions. In the tumor xenografts, the distribution of TK/eGFP and CD/UPRT/mDsRed expression visualized with fluorescence microscopy was co-localized with the hypoxia marker pimonidazole positive staining cells. Furthermore, administration of 5-FC and GCV in mice in combination with local irradiation resulted in tumor regression, as compared with prodrug or radiation treatments alone. Our data suggest that the hypoxia-inducible TK/GCV+CDUPRT/5-FC triple suicide gene therapy may have the ability to specifically target hypoxic cancer cells and significantly improve the tumor control in combination with radiotherapy.


Introduction
Hypoxia is a major determinant of the malignant progression of tumor cells and their response to therapy �1�. Clinical studies have shown a strong correlation between tumor pO 2 , local control and overall survival �2,3�. Since the vast majority of human solid tumors have median pO 2 levels lower than that of the surrounding normal tissues �4�, efforts have been made to exploit it as a target for cancer treatments. At the molecular level, hypoxia is a powerful modulator of gene expression. An important mediator of hypoxic response is the interaction of a transcriptional complex termed hypoxia inducible factor �HIF� with its cognate DNA recognition site, the hypoxiaresponse element �HRE� ��,6�. The HRE sequences isolated from oxygen-responsive genes have been shown to selectively induce gene expression in response to hypoxia when placed upstream of a promoter, and hypoxia-targeted gene therapy can be achieved �7,8�.
Two well-established gene-directed enzyme prodrug therapy strategies, in which Herpes simplex virus 1-thymidine kinase �HS�1-TK� or cytosine deaminase �CD� was used in combination with prodrug ganciclovir ��C�� or �-�uorocytosine ��-FC� respectively, have been proposed for hypoxia-driven gene therapy in different tumor models �9,10�. Double suicide gene combined with HS�1-TK/�C� and CD/�-FC resulted in enhanced anticancer effects and has been used in clinical trials �11,12�. However, hypoxia-targeted double suicide gene therapy has not been reported. Studies from us and other groups have demonstrated that co-transduction of uracil phosphoribosyltransferase �UPRT� with CD gene to cancer cells improved antitumor efficacy �13,14�. Furthermore, our studies demonstrated that hypoxia-targeted expression of bifunctional suicide gene CDUPRT enhanced the cytotoxicity of �-FC treatments and radiosensitization under hypoxic conditions �1��. We also found the superior antitumor effects of triple suicide gene approach-TKCDUPRT over double gene approach-TKCD in a rat prostate tumor model �16�.
In the present study, we generated a human colorectal cancer model expressing triple suicide gene �HS�1-TK, CD and UPRT� under the control of the hypoxia inducible promoter. Significant radiosensitization effects were demonstrated both in vitro and in vivo with �C� and �-FC treatments. Our data suggest that the hypoxia-inducible triple suicide gene therapy has the ability to specifically target hypoxic cancer cells and significantly improves the tumor control in combination with radiotherapy. This novel model also appears to be a valuable tool to study tumor hypoxia, radiation effects and hypoxiatargeted radio-gene therapy.
HCT-8 cells were co-transfected with two plasmids; p9HRE-TK/e�FP, which contains the TK/e�FP fusion gene under the control of a hypoxia-inducible promoter and a constitutively expressed neomycin-resistance gene �Neo r � �17�, and p9HRE-CD/UPRT/mDsRed, which contains the CD, UPRT and mDsRed fusion gene under the regulation of the hypoxiainducible promoter and a hygromycin B-resistance gene �Hyg r � �1�� �Fig. 1A�. The hypoxia-inducible promoter consists of nine tandem repeats of the HRE from human erythropoietin gene linked to the S�40 minimal promoter �S�40min�. The stable cell line expressing HRE-regulated TK/e�FP and CD/UPRT/mDsRed was established by �uorescence-activated cell sorting using the cell sorter �MoFlo; Dako, Carpinteria, CA, USA� �17� and designated as HCT8-HRE cells. HCT-8 cells transfected with empty vector were used as the control cells.
Western blot analysis. The cells were incubated under normoxic or 0.�% O 2 conditions for 12, 24, 48 or 72 h, and then whole cell extracts were prepared. The protein was detected with the sheep anti-yCD polyclonal antibody �Biotrend Chemikalien �mbH, Cologne, �ermany� and the anti-HS�1-TK monoclonal antibody �kindly provided by Dr W.C. Summers, Yale University, New Haven, CT, USA�.
Flow cytometry. After incubation under normoxic or 0.�% O 2 conditions for 48 h, the cells were trypsinized, centrifuged and re-suspended in PBS at 4˚C. The �uorescence of e�FP and mDsRed was measured using the cell sorter �MoFlo�. The data were analyzed with the FlowJo program �Tree Star Inc., Ashland, OR, USA�. Colony formation assay. The cells were treated with �-FC �Invivo�en, San Diego, CA, USA� or �C� �Sigma-Aldrich, St. Louis, MO, USA� at various concentrations under normoxic or 0.�% O 2 conditions for 48 h. The cells were then trypsinized, counted, serially diluted and plated into 60-mm dishes. Clonogenic survival was determined by counting crystal violet-stained colonies ~14 days later. The survival fraction was normalized to the cell survival without drug treatments and plotted as a function of the drug concentration.
MTT assay. The cells were plated into 96-well plates �10 4 /well� and incubated with various concentrations of �C� and/or �-FC for 48 h under normoxic or 0.�% O 2 conditions. The medium containing the drug was then replaced with fresh medium, and cells were cultured for an additional 48 h under the normoxic conditions. Tetrazolium dye �Sigma� was added and allowed to react for 2 h at 37˚C. After dissolving in DMSO, dye conversion was read using an ELISA plate reader at 490 nm against 610 nm.
Bystander effect. The HCT-8 control cells were mixed with HCT8-HRE cells at the 80:20 ratio and plated into 96-well plates. The cells were treated with various concentrations of �C� and/or �-FC for 48 h under 0.�% O 2 conditions, and cell viability was determined with the MTT assay described above.
Radiation survival assay. The cells were exposed to �-FC �0.� mg/ml� and �C� �� µg/ml� under normoxic or hypoxic conditions for 48 h, and then irradiated for various doses using a Cs-137 unit �Mark 1 model 68; J L Shepherd & Associates, San Fernando CA, USA� at ~2.0 �y/min. Clonogenic survival was then determined by colony formation assay. The radiation survival fractions were normalized for plating efficiency using non-irradiated cells with or without drug treatment and plotted as a function of radiation doses.
Xenograft model. Animal protocols were approved by the Institutional Animal Care and Use Committee. Xenografts were formed by injecting �x10 6 cells subcutaneously into the hind legs of 6-8-week old female nude mice �athymic nu/nu; NCI Frederick Cancer Research Institute, Frederick, MD, USA�. When the xenografts reached ~10 mm in diameter, the mice were injected i.v. with pimonidazole �80 mg/kg; HPI, MA�. Two hours later, Hoechst 33342 �2� mg/kg; Sigma-Aldrich� was i.v. injected, and 1 min later, the mice were sacrificed by CO 2 breathing. Then, tumors were excised and snap-frozen in OCT mounting medium �Sakura Finetek USA Inc., Torrance, CA, USA�. Frozen sections of 8-µm thickness were prepared for the immunohistochemical analysis.
Immunohistochemical staining and fluorescent microscopic imaging. Tumor sections were first imaged for the e�FP, mDsRed and Hoechst 33342 �uorescence. They were then stained for pimonidazole using FITC-conjugated murine antipimonidazole monoclonal antibody �17�. Finally, the sections were stained with hematoxylin-eosin with the standard protocol. Fluorescence images were acquired at x�0 magnification using the �uorescence microscope �Axiovert 200 M� equipped with a CCD digital camera, a computer-controlled motorized stage and MetaMorph 7.0 Imaging software �Molecular Devices, Sunnyvale, CA, USA�.
Tumor growth delay. Each tumor was measured with digital caliper in three orthogonal dimensions �a, b and c�, and tumor volume was calculated as πabc/6. When the tumor grew to ~100-1�0 mm 3 , the mice were divided into four groups �10 mice/group� and treated with: i� PBS, ii� �-FC �i.p. �00 mg/kg� + �C� �i.p. 30 mg/kg� daily for 14 days, iii� radia-/kg� + �C� �i.p. 30 mg/kg� daily for 14 days, iii� radia-kg� + �C� �i.p. 30 mg/kg� daily for 14 days, iii� radiation �1� �y� on the �th day, iv� �-FC+�C� daily for 14 days followed by radiation �1� �y� on the �th day, respectively. The tumor volume was measured three times a week until the volume exceeded 1,000 mm 3 and plotted as a function of time.
Statistical analysis. Averages are presented as the mean ± SE. Difference in cytotoxicity and radiosensitivity was determined using the Student's t-test. A P-value of <0.01 was considered to indicate a statistically significant difference.

Results
Hypoxia-induced expression of triple suicide gene. HCT-8 stable cell lines expressing HRE driven TK/e�FP and CD/UPRT/mDsRed were successfully established. Several stable transfectants were established and their gene expressions were similar as verified by western blot analysis, �ow cytometry, hence one of them was used for further experiments. Western blot analysis showed that the expression level of TK/e�FP and CD/UPRT/mDsRed protein was markedly increased after hypoxic treatments in a time-dependent manner �Fig. 1B�. Exposure of the HRE-TKCDUPRT cells to hypoxia led to an increase of the hypoxia-regulated e�FP and mDsRed �uorescence. In �ow cytometric analysis, an 80-fold increase in e�FP and 4-fold increase in mDsRed �uorescence was observed in cells treated at 0.�% O 2 for 48-h hypoxia compared to normoxic control cells �Fig. 1C�. The �uorescent microscopic images of hypoxic and normoxic cell showed that strong e�FP and mDsRed �uorescence signal was observed only in hypoxic cells, confirming the results of �ow cytometric analysis �Fig. 1D�.

Hypoxia-induced TK/CD/UPRT expression sensitizes HCT8-HRE cells to 5-FC and GCV.
It was clearly shown that the cell clonogenic survival was significantly inhibited by �C� �Fig. 2A� or �-FC �Fig. 2B� under hypoxic conditions compared to normoxic conditions. Whereas HCT8-HRE cells were sensitive to �C� and �-FC administered independently, concurrent prodrug treatment resulted in much greater cytotoxicity �Fig. 2C�. Another method, MTT assay, was used to further assess hypoxia-induced cytotoxicity to �C� and �-FC. It was found that cell viability under hypoxic conditions was significantly inhibited by �C� or �-FC, relative to that under normoxic conditions �Fig. 3�. It was demonstrated again that concurrent prodrug treatment resulted in much greater cytotoxicity. Fig. 4. Treatments with �C� plus �-FC had little radiosensitizing effect in normoxic HCT-HRE cells. However, drug treatments did enhance the radiosensitivity of hypoxic HCT-HRE cells. effects. Bystander killing was examined in vitro by culturing mixture with 20% HCT8-HRE cells and 80% control cells and exposing the cells to increasing doses of �C� and/or �-FC under the hypoxic conditions. As shown in Fig. �, the cell mixture of 20% HCT8-HRE cells was much less sensitive to �C� than 100% HCT8-HRE cells under the hypoxic conditions �Fig. �A�, whereas 20% HCT8-HRE cell mixtures was slightly less sensitive to �-FC than 100% HRE cells �Fig. �B�. Furthermore, combined �C� and �-FC treatments also led to greater cytotoxicity in the mixed cell population �Fig. �C�.

Hypoxia-induced transgene expression in vivo.
Hypoxiadriven transgene expression in vivo was validated and compared with distributions of exogenous hypoxia marker �pimonidazole� and blood perfusion marker �Hoechst 33342�. Fig. 6 provides a detailed examination and comparison of the different biomarkers in magnified views of a region from an HCT8-HRE tumor. The merged images of e�FP and Hoechst �Fig. 6A� or mDsRed and Hoechst �Fig. 6B� show that e�FP �green� and mDsRed �red� are located between well perfused area �blue� and the necrotic region, consistent with hypoxiainduced expression of the e�FP and mDsRed reporter gene. Fig. 6C shows that the exogenous hypoxic marker pimonidazole �green� is also accumulated in regions of low perfusion.
Radiation combined with 5-FC and GCV treatments delays tumor growth. The response of the HCT8-HRE tumors to prodrug ��C� and �-FC� treatment with or without radiation was assessed using the tumor growth delay assay. As shown in Fig. 7, �-FC ��00 mg/kg� and �C� �30 mg/kg� treatments for 14 days alone or radiation �1� �y� alone, slightly delayed  the tumor growth. However, treatments of �C� plus �-FC combined with irradiation showed significant tumor growth delay �P<0.01�.

Discussion
The hypoxic microenvironment, an important feature of human solid tumors but absent in normal tissue, may provide an opportunity for cancer-specific gene therapy. Studies have demonstrated the potential of hypoxia/HRE-regulated gene therapy for cancer with HS�1-TK �10,18�, bacterial CD �7,19,20� and yeast CD �21,22� employed. To explore the possibility of the hypoxia-targeted triple suicide gene therapy strategy,    HCT8-HRE cells were exposed to �-FC �0.� mg/ml� and �C� �� µg/ml� under normoxic or hypoxic conditions for 48 h, followed by various doses of radiation. Surviving fractions were determined using colony formation assay. Each data point and error bar represents average and SE of three independent experiments. IR, irradiation.
One potential advantage of suicide gene therapy relevant to clinical application is its radiosensitization effect. Our data clearly showed that the co-expression of hypoxia-regulated HS�1-TK, CD and UPRT triple genes not only significantly increased the sensitivity of �C� and �-FC, but also improved the radiosensitizing effect of �C� and �-FC in vitro �Fig. 4� and in vivo �Fig. 7�. It has been shown that HS�-1 TK/�C� gene therapy may inhibit the repair of radiation-induced sublethal DNA damage �23�. Others have suggested that the radiosensitization in CD/�-FC and CD/UPRT/�-FC approaches is mediated through the inhibition of thymidylate synthase by �-FdUMP, resulting in the depletion of deoxythymidine monophosphate pools and increased DNA strand break, as well as redistribution of cells to the radiosensitive early S phase �24�. Previously, we demonstrated that the CDUPRT/�-FC approach has greater radiosensitization effect than the CD/�-FC system �13�. The hypoxia-regulated CD/�-FC approach has shown significant tumor controlling effects as combined with irradation in different tumor models �19,22�. Therefore, it is not surprising that significant tumor control effects were found as the triple suicide gene approach was combined with radiation.
Another advantage of suicide gene therapy approaches is the bystander effects. The HS�1-TK/�C� system requires direct cell-to-cell contacts and active transport of toxic �C� metabolites to neighboring cells through the gap junctions to exert their bystander killing effect �2��. On the other hand, in the CD/�-FC system, the converted �-FU is able to diffuse across the cell membrane into adjacent cells without going through the gap junction, resulting in a more powerful bystander effect �26�. Strong bystander effect has also been shown with the CDUPRT/�-FC approach �27�. In the present study, the cell mixture of 20% HRE-TKCDUPRT cells was more sensitive to �-FC than �C� under the hypoxic conditions �Fig. ��, whereas the sensitivity of the 100% HRE cells to �C� and �-FC was similar as determined by MTT assay �Fig. 3�. This is partly due to the low expression of gap junctions in the human colorectal cells �2��. Even so, combined �C� and �-FC treatments leads to greater cytotoxicity in mixed cell populations �Fig. �C�, which demonstrated the superior bystander killing in the hypoxia-driven TKCDUPRT approach.
Several factors may impact on the efficacy of hypoxiatargeted gene therapy. Firstly, the tumors with high hypoxia fraction, for example large tumors, will be targets for hypoxiaregulated suicide gene therapy. In the present study, treatment of �-FC plus �C� in tumor-bearing mice showed only mild tumor growth delay �Fig. 7�. It may be due to low hypoxia fraction in the model used �Fig 6�. Secondly, HIF activity in tumors depends on availability of the HIF-1α/2α subunit, the levels of which increase under hypoxic conditions and through activation of oncogenes and/or inactivation of tumor suppressor genes �such as p�3�. HCT-8 tumor cells bear wild-type p�3 gene, which may limit the hypoxia-induced gene expression in vivo. To investigate the potential role of hypoxia-targeted triple suicide gene therapy, tumor models with different hypoxia conditions and gene profile were used.
The potential limitation of the present study is the use of stably transfected cell lines and associated tumor models. As two plasmids carrying hypoxia-regulated TK or CDUPRT were used, the hypoxia-induced expression/function and cytotoxicity of TK and CDUPRT were not comparable �Figs. 1 and 2�. To translate this strategy to clinical application, we are generating adenovirus vectors to deliver hypoxia-regulated TK, CD and UPRT genes into in vivo tumor models and to test whether triple gene transduction following �C� and �-FC treatment will improve the efficacy of radiotherapy. To overcome the limitations in introducing gene therapy into clinical use, advanced methods of hypoxia-targeted gene delivery are also being developed.
In conclusion, our data suggest that the hypoxia-inducible TK/�C�+CDUPRT/�-FC triple suicide gene therapy may have the ability to specifically target hypoxic cancer cells and improve the tumor control in combination with radiotherapy. This novel model may also serve as a valuable tool to study the correlations among tumor hypoxia, radiation and hypoxiatargeted radio-gene therapy.