Abstract
Double prodrug activation gene therapy using the Escherichia coli cytosine deaminase (CD)herpes simplex virus type 1 thymidine kinase (HSV1-tk) fusion gene (CD/TK) with 5-fluorocytosine (5FC), ganciclovir (GCV), and radiotherapy is currently under evaluation for treatment of different tumors. We assessed the efficacy of noninvasive imaging with [124I]FIAU (2′-fluoro-2′-deoxy-1-β-d-arabinofuranosyl-5-iodo-uracil) and positron emission tomography (PET) for monitoring expression of the CD/TK fusion gene. Walker-256 tumor cells were transduced with a retroviral vector bearing the CD/TK gene (W256CD/TK cells). The activity of HSV1-TK and CD subunits of the CD/TK gene product was assessed in different single cell-derived clones of W256CD/TK cells using the FIAU radiotracer accumulation assay in cells and a CD enzyme assay in cell homogenates, respectively. A linear relationship was observed between the levels of CD and HSV1-tk subunit expression in corresponding clones in vitro over a wide range of CD/TK expression levels. Several clones of W256CD/TK cells with significantly different levels of CD/TK expression were selected and used to produce multiple subcutaneous tumors in rats. PET imaging of HSV1-TK subunit activity with [124I]FIAU was performed on these animals and demonstrated that different levels of CD/TK expression in subcutaneous W256CD/TK tumors can be imaged quantitatively. CD expression in subcutaneous tumor sample homogenates was measured using a CD enzyme assay. A comparison of CD and HSV1-TK subunit enzymatic activity of the CD/TK fusion protein in vivo showed a significant correlation. Knowing this relationship, the parametric images of CD subunit activity were generated. Imaging with [124I]FIAU and PET could provide pre- and posttreatment assessments of CD/TK-based double prodrug activation in clinical gene therapy trials.
Introduction
Prodrug activation or so-called “suicide” gene therapy has been shown to be successful in potentiating the therapeutic index by sensitizing genetically modified tumor cells to various prodrugs or enhancing the action of commonly used chemotherapeutic agents. During the past decade, several prodrug activation and sensitization approaches have been tested in different tumor models in animals [1–3]. Two of the most studied prodrug activation approaches involve transfection of tumors with herpes simplex virus type 1 thymidine kinase (HSV1-tk) gene or Escherichia coli cytosine deaminase (CD) genes followed by administration of ganciclovir (GCV) or 5-fluorocytosine (5FC), respectively [4–19].
HSV1-TK phosphorylates a nontoxic acycloguanosine analogue, GCV to GCV-MP, which is subsequently converted to GCV-diphosphate and GCV-triphosphate by endogenous di- and triphosphate kinases and is incorporated into DNA of proliferating cells by DNA-polymerase. Because GCV does not have a functional 3′-hydroxyl group, upon incorporation into DNA it acts as a DNA chain terminator and triggers cell apoptosis [15–17]. Surrounding cells that do not express HSV1-tk may share to some extent the phosphorylated GCV through intercellular gap junctions and also undergo apoptosis [18–22]. The latter constitutes the so-called “bystander” effect, which has been used as a justification for tumor gene therapy, despite the relatively low efficacies of in vivo gene delivery vectors.
CD deaminates nontoxic 5FC to 5-fluorouracil (5FU) [23],[24]; upon sequential ribosylation by uridine phosphorylase and phosphorylation by uridine kinases to 5FUMP, 5FUDP, and 5FUTP, which inhibit RNA synthesis [25]. 5FU is also metabolized to 5FUdMP, 5FUdDP, and 5FUdTP that irreversibly inhibit thymidylate synthase (TS) and cause depletion of the thymidine (TdR) pool in proliferating cells, inhibit cell proliferation, and cause apoptosis [26–28]. The magnitude of the bystander effect in the CD/5FC approach is significantly larger than that in HSV1-tk/GCV, because the 5FU produced by the deamination of 5FC can easily cross cell membranes of transduced and non-transduced cells and does not require cell junctions [29].
The majority of studies in animals using adenoviral-mediated in vivo transfer of HSV1-tk or CD genes coupled with GCV and CD therapies, respectively, have demonstrated marked antitumor effects [14,30–33]. Numerous ongoing clinical trials that followed these promising preclinical results sought to determine the efficacy of HSV1-tk/GCV and CD/5FC therapies in humans. However, despite the established efficacy of these approaches in animal models, many studies failed to achieve complete tumor regressions with either the HSV1-tk/ GCV [8,34–36] or CD/5FC therapies [9,11,13,14]. To improve the efficacy of prodrug activation gene therapy of tumors, several studies assessed the combination of different approaches with or without external beam radiation therapy. A pronounced antitumor effect and tumor radiosensitization was demonstrated by several studies that combined HSV1-tk/ GCV and CD/5FC therapies [37–42]. The potentiation effect of combined HSV1-tk/GCV and CD/5FC therapies is mediated through different cytotoxic mechanisms, and their radiosensitization effects target different cell populations [37,39–42]. To optimize and simplify the delivery of HSV1-tk and CD genes to tumor cells, a cytosine deaminase-thymidine kinase fusion gene (CDglyTK) was developed in which the CD and HSV1-tk cDNAs are linked in tandem by a polyglycine linker coding DNA sequence [37]. Combined CDglyTK-GCV/5FC/radiation therapy is currently undergoing clinical trials.
Noninvasive imaging would significantly aid in optimization and management of the combined CDglyTK-GCV/5FC/radiation therapy by defining the location, magnitude, and persistence of CDglyTK gene expression in target tumor tissues. Radionuclide imaging of CD expression in vivo using radiolabeled 5FC is not feasible, because the CD/5FC reaction product, radiolabeled 5FU, rapidly crosses the cell membrane and does not effectively accumulate in CD-expressing cells [43]. However, it has been recently demonstrated that noninvasive quantitation of CD expression in transduced tumors is feasible using NMR spectroscopy during therapy with 5FC [44]. It should be possible to use NMR spectroscopy to monitor CD/TK expression in transfected tumors; however, current levels of sensitivity and resolution of NMR spectroscopic imaging are comparatively poor [45]. Therefore, we have developed and validated an “indirect” nuclear imaging paradigm to image the expression of the CDglyTK fusion gene in transduced tumors.
In a series of previous studies, we demonstrated that HSV1-tk gene expression can be imaged with 131I or 124I-labeled 2′-fluoro-2′-eoxy-β-d-arabinofuranosyl-5-iodo-uracil (FIAU) and a gamma camera [46],[47], SPECT [46], or positron emission tomography (PET) [48], respectively. Similar results have been demonstrated by others using 18F-labeled acycloguanosine analogues [49–51]. More recently, we demonstrated that the HSV1-tk gene could be used as a reporter gene to monitor and quantitatively assess the expression of a second gene that is linked to the reporter gene by an internal ribosome entry site (IRES) sequence [52], and this was subsequently confirmed by others [53].
An indirect method for noninvasive imaging of therapeutic gene expression using a reporter gene (e.g., HSV1-tk) requires constant and proportional coexpression of these genes over a wide range of expression levels. Strictly equimolar coexpression of two proteins can be achieved by generating a fusion protein, such as CDglyTK. In the current study, we assessed the feasibility and efficacy of imaging the expression of CDglyTK fusion gene with [124I]FIAU and PET. In this imaging paradigm, the HSV1-TK subunit of the CDglyTK fusion gene has a dual function — as a therapeutic gene and as a reporter gene for monitoring and quantitation of both HSV1-TK and CD subunit activity. We show that CDglyTK gene expression in W256CD/TK subcutaneous tumor xenografts in rats can be monitored by [124I]FIAU and PET imaging in vivo. We demonstrate that PET imaging is sufficiently sensitive to discriminate different levels of CDglyTK gene expression, and that parametric images of CD subunit activity can be generated.
Materials and Methods
Cell Cultures and Retroviral Vector
Walker-256 rat mammary carcinoma cells (W256) were obtained from American Type Tissue Culture Collection and grown as monolayers in Eagle's medium (MEM) supplemented with 10% FCS. The pWZLneoCDglyTK recombinant retroviral vector (Figure 1) was kindly provided by Dr. Kenneth Rogulski (Henry Ford Hospital, Detroit, MI, USA). This vector encodes for E. coli CD and HSV1-TK fusion protein, in which the two cDNAs encoding the corresponding enzymes are linked in tandem by an 11-amino acid linker consisting of 10 glycines and a single phenylalanine (CDglyTK) as described previously [37]. The pWZLneoCDglyTK is a single promoter vector, having all gene expression under control of the Moloney murine leukemia virus long terminal repeat (LTR). Inclusion of the IRES from the encephalomyocarditis virus [55] into the expression cassette allows for neomycin-resistance gene translation from the same bicistronic proviral mRNA. The gp-E86-based pWZLneoCDglyTK retroviral vector producer cells were grown as monolayers in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FCS. The in vitro transduction of W256 cells with the pWZLneoCDglyTK vector was accomplished by exposing the W256 cell monolayers to filtered (0.45 μm) supernatant obtained from the vector producer cells. Forty-eight hours posttransfection, W256CDTK cells were selected in a medium containing 800 μg/ml of geneticin (G418). Following 7 days of G418 selection, multiple single cell-derived clones of W256CDTK cells were isolated and grown in MEM supplemented with 10% FCS and 250 μg/ml G418. These stably transduced W256CDTK cell clones were characterized for levels of CDglyTK transgene expression as described below. Single cell-derived clones of W256CDTK cells with significantly different levels of HSV1-tk expression were selected and used in further studies.
Schematic structure of the pWZLneoCDglyTK retroviral vector, LTR, spicing donor (SD), splicing acceptor (SA), and packaging signal.
FIAU Accumulation Assay
FIAU accumulation assays were performed as previously described [56]. Briefly, different single cell-derived clones of W256CDTK cells were seeded in 150 × 25 mm tissue culture plates at a concentration of 1 × 106 cells/plate and grown until 50–60% confluent. The incubation medium was replaced with 14 ml of medium containing [2-14C]FIAU 0.01 μCi/ml (56 mCi/mmol) and [met-3H]TdR 0.2 μCi/ml (60 Ci/mmol) (Moravek Biochemicals, CA). Radiochemical purity of each compound was checked in our laboratory using high-performance liquid chromatography and found to be >98%. The cells were harvested by scraping after various periods of incubation (10, 30, 60, 90, and 120 min), centrifuged, cell pellets weighed, and assayed for radioactivity concentration (dpm/g) using a TriCarb 1600 beta spectrometer (Packard, CT) with standard wn>3H and 14C dual channel counting techniques. The medium was also counted before and after incubation. The data were expressed as a harvested cell-to-medium concentration ratio (dpm/g cells)/(dpm/ml medium). The rates of accumulation (Ki) for FIAU and TdR were determined from the slope of the cell-to-medium ratios versus incubation time plots and have units of tracer clearance from the medium (ml medium/min/g cells). The ratio of Ki values (the FIAU/TdR ratio) is a measure of HSV1-TK activity and correlates with independent measures of the gene expression [56].
CD Enzyme Assay
An in vitro assay for measuring CD activity was performed as described previously [12] with minor modifications. Each clone of W256CDTK cells was seeded into tissue culture plates at a concentration of 1 × 107 cells/plate. Upon reaching 60% confluence, the cells were harvested by scraping into sterile PBS. Cell pellets were obtained by centrifugation (3000 rpm for 7 min at 37°C). The supernatant was aspirated and the cell pellet weighed. The pellet was resuspended in 800 μl of buffer (100 mM Tris-Cl pH 8.0, 1 mM EDTA, and 1 mM β-mercaptoethanol), lysed with ultrasonication, and centrifuged for 10 min. Cell extract (500 μl) was transferred into a 1.5-ml vial and 100 μl of 3 mM [3H]cytosine (Moravek Biochemicals) was added to the cell extract. The reaction mixture was incubated at 37°C and multiple samples (60 μl each) were obtained from the reaction mixture every 30 sec for 5 min. The reaction was terminated by adding the sampled aliquots to 300 μl of 0.1 M acetic acid. Then, the acidified samples were loaded onto a strong cation exchange column, Bond Elut SCX (Varian, CA); acetic acid (0.1 M, 3 ml) was used to elute [3H]uracil and potassium phosphate (0.1 M, 3 ml) was used for [3H]cytosine elution. Aliquots (500 μl) from the eluted fractions containing [3H]uracil and [3H]cytosine, respectively, were obtained in triplicate for each time point and the radioactivity was measured by beta spectroscopy (Tri-Carb, Packard). The enzymatic rate constant was determined from the radioactivity/time plots (converted to μmol substrate/min/g) and expressed as units of enzyme activity per gram cells (U/g cells) [12].
Subcutaneous Tumors
The experimental protocol involving animals was approved by the Institutional Animal Care and Use Committee of the Memorial Sloan Kettering Cancer Center. Six subcutaneous tumors were produced in each animal. Five different clones of W256CDglyTK cells (same as in in vitro studies) were inoculated subcutaneously (106 cells in 100 μl of serum-free MEM) into rnu-rnu rats (Harlan Sprague-Dawley, Indianapolis, IN) weighing 200–250 g, at different sites, including the dorsal part of both shoulders, both thighs, and the right flank. A wild-type W256 tumor was produced in the left flank and served as a negative control. Tumor growth was monitored by daily volume measurements. The imaging studies were conducted when the tumor size reached a mean of 1 cm3.
No Carrier-Added Synthesis [124I]FIAU
No carrier-added 124I was produced (p,n) from an enriched tellurium-124 oxide target, which after irradiation was subjected to dry distillation to release the trapped radioiodide [57]. The 124I was isolated in a small volume of phosphate buffer. [124I]FIAU was prepared from the stannylated precursor, 2′-fluoro-2′-deoxy-1-β-d-arabinofuranosyl-5-(tri-n-butyl-tin)-uracil [58], which was allowed to react with [124I]NaI in the presence of a mixture of acetic acid/30% hydrogen peroxide. After quenching with sodium metabisulfite, the [124I]FIAU was isolated on a C-18 Sep-Pak cartridge system, then eluted off with methanol. After evaporation of the methanol, the product was reconstituted in sterile, pyrogen-free, physiological saline (with 5% ethanol added) and passed through a sterile nonpyrogenic 0.22 μm Millipore filter. The radiochemical purity was determined by radio-TLC (Sigma-Aldrich silica gel plates, eluent: ethyl acetate/acetone/water, 14:8:1) and was found to be > 95%.
Positron Emission Tomography
No carrier-added [124I]FIAU (25 μCi per animal) was injected intravenously 14 days after subcutaneous inoculation of W256CDglyTK tumor cells. One day prior to [124I]FIAU administration, all animals received a 2-ml intraperitoneal injection of a 0.9% NaI solution to block thyroid uptake of radioactive iodide, small quantities of which result from the radiolysis of [124I]FIAU. PET images of [124I]FIAU accumulation were obtained at 24 hr after tracer administration. PET was performed using an Advance PET tomograph (General Electric, Milwaukee, WI). Transmission and emission scans were obtained in 3-D mode; the emission scans were corrected for randoms, dead time, and attenuation. Duration of the transmission scans was 10 min and the duration of the emission scans was 40 min. Images were reconstructed with filtered backprojection (Hanning filter with cutoff frequency of 0.5 cycles/pixel), yielding a slice thickness of 4.2 mm. A 124I reference standard of known radioactivity concentration was placed within the field of view of the PET scanner and used for quantitation of the images.
Functional coexpression of the CD and HSV1-tk subunits of CDglyTK transgene product in different single cell-derived clones of W256CDTK+ cells in vitro. Enzymatic activities for the CD and HSV1-tk subunits of the CD/TK fusion protein (expressed as U/g cells and FIAU/TdR accumulation ratio, respectively) are plotted for different clones. Each point represents a single clone, and represents the average ± SD of triplicate studies. A linear relationship is observed (r = .99).
Assessment of [124I]FIAU Accumulation and CD Expression in Tumors
After PET imaging, the animals were sacrificed, the tumors sampled, weighed, and directly assayed for radioactivity using 5500 Packard gamma spectrometer (Packard). A part of each tumor sample was homogenized in a cell lysis buffer and CD activity was assayed as described above. The levels of HSV1-tk expression ([124I]FIAU %dose/g) and CD (enzyme U/g) gene expression in corresponding tumor samples were compared.
Statistics
Descriptive statistics of group data were performed using univariate analysis. Group data were compared using ANOVA analysis, regression analysis, and Student's t test; a p value of #x003C;.05 was considered significant. Statistical analysis of data was performed using the StatView 4.57 (Abacus Concepts, CA).,/p>
Results
Relationship Between the HSV1-TK Expression (FIAU Accumulation) and CD Expression In Vitro
Five clonal populations of W256CDTK tumor cells (F6, E1, C3, J1, and E3) were characterized for the level of CDglyTK fusion gene expression. The levels of CD and HSV1-tk enzyme activity were determined in each of the clones and the results demonstrate that CD and HSV1-tk subunit coexpression is tightly coupled. A plot of CD and HSV1-TK enzyme activity in these clones in Figure 2 shows a highly significant linear relationship (p #x003C; .05, <i>t = 2.86, Z = 3.66; paired t test).
Imaging CDglyTK transgene expression in subcutaneous W256CDTK+ tumors with [124I]FIAU and PET. Axial PET images from two animals are shown. The location of subcutaneous tumors grown from single cell-derived clones of transduced W256CDTK+ is shown in the middle panel. The images are color-coded and scaled to a color intensity bar; the intensity bar shows the range of [124I]FIAU accumulation (%dose/g) that reflects the activity of the HSV1-tk subunit, and the range of CD enzyme activity (U/g tissue) of the CD subunit based on the linear relationship shown in Figure 4.
PET Imaging of CDglyTK Fusion Gene Expression in Subcutaneous W256CDTK Tumors
To assess whether [124I]FIAU PET imaging of transduced W256CDTK tumors is sufficiently sensitive to detect and discriminate between different levels of CDglyTK transgene expression, we studied nine animals. [124I]FIAU-derived radioactivity was localized to the areas of W256CDTK tumors (Figure 3). The levels of accumulated radioactivity varied in different W256CDTK tumors, although a consistent pattern was observed (WCDTK F6>E1>C3>J1>E3>wild-type W256). Radioactivity (%dose/g) in tumors, muscle, and blood plasma was also assessed by tissue sampling and by gamma spectroscopy after the animals were killed to confirm the validity of the PET measurements (Table 1). The range of radioactivity retained in different clonal W256CDTK tumors was from 0.018 ± 0.005 %dose/g in the lowest expressing clone to 0.73 ± 0.16 %dose/g in the highest expressing clone. Wild-type W256 tumors had substantially less accumulation (mean 0.011 ± 0.006 %dose/g) of FIAU-derived radioactivity.
Relationship Between the HSV1-tk Expression (FIAU Accumulation) and CD Enzyme Activity in Transduced Tumors
To assess whether PET imaging of HSV1-tk gene expression in transduced tumors reflects the level of CD coexpression, we compared the levels of [124I]FIAU accumulation (%dose/g) and the levels of CD enzyme activity (U/g tumor) obtained from the same samples of W256CDTK tumor tissue. A highly significant linear relationship (p #x003C; .0001, <i>t = 19.5, Z = 9.11; paired t test) was observed between the levels of [124I]FIAU accumulation (%dose/g) and the levels of CD enzyme activity (U/g tumor) (Figure 4). The slope of the relationship in Figure 4 constitutes a sensitivity index for the FIAU-PET image, and this slope can be used to generate a color-coded scale bar for a parametric image that reflects CD subunit activity in transduced tumors (Figure 3).
The Average Levels of CD and HSV1-tk Subunits of CDglyTK Transgene Product in Different Subcutaneous W256CDTK+ Tumors In Vivo.
Functional coexpression of the CD and HSV1-TK subunits of CDglyTK transgene product in different subcutaneous W256CDTK+ tumors in vivo. Enzymatic activity for the CD and HSV1-tk subunits of the CD/TK fusion protein was assessed in the same tissue samples obtained from different W256CDTK+ subcutaneous tumors (as shown in Figure 3). CD (U/g tumor) and HSV1-TK (FIAU %dose/g tumor) enzyme activity are plotted for each individual tissue sample; a linear relationship is observed (r = .92).
Discussion
In the current study, we demonstrated that noninvasive imaging of CDglyTK expression in stably transduced W256CDTK subcutaneous tumor xenografts in rats is feasible with [124I]FIAU and a clinical PET tomograph. Different levels of CDglyTK gene expression could be easily discriminated by PET imaging of subcutaneous tumor xenografts grown from single cell-derived clones of stably transduced W256CDTK cells expressing different levels of the CDglyTK gene. Multiple subcutaneous tumors in the same animal expressing different levels of CDglyTK can be viewed as a model of heterogeneous gene expression within a single tumor after in vivo gene transfer. Such a multi-tumor model provides for more statistically significant comparisons between transduced tumors expressing different levels of the CDglyTK gene and wild-type tumor, because all tumors are exposed to the same levels of radiotracer in blood (same input function).
These results are in agreement with our previous studies using a similar multi-tumor model, in which we demonstrated that the levels of HSV1-tk expression in transduced subcutaneous tumors measured by [124I]FIAU and PET (%dose/g tissue) correlate with other independent measures of HSV1-TK expression (e.g., mRNA levels and sensitivity to GCV) [48]. A good correlation was also observed between different levels of [131I]FIAU accumulation (%dose/g tissue) imaged by a gamma camera and corresponding levels of HSV1-tk expression following intratumoral administration of different doses of adenoviral vectors bearing the HSV1-tk gene in a mouse model of colon carcinoma metastasis to the liver [47]. Using a p53-responsive inducible reporter system, we recently demonstrated that the levels of [124I]FIAU accumulation measuredby PET correlate with both the levels of HSV1-tk mRNA and HSV1-TK protein as measured in tumor tissue samples by RT-PCR and Western blot, respectively [54]. These combined data demonstrate that HSV1-tk gene imaging with [*I]FIAU and nuclear imaging techniques is a sensitive, quantitative, and reliable method for reporter gene imaging, and that it can be used to image and measure the level of CDglyTK transgene expression.
The levels of HSV1-TK subunit activity linearly correlated with the levels of CD subunit activity in corresponding clones of W256CDTK cells in vitro as well as in subcutaneous tumor xenografts in vivo over a log order wide range of CDglyTK expression levels. A strong correlation between the activities of CD and HSV1-TK subunits observed in this study was expected because they are expressed at equimolar levels as a single chimeric protein. Previously, we demonstrated that the activity of two subunits of a fusion protein produced by a fusion gene (thymidine kinase-green fluorescent protein fusion gene, TKGFP) are highly correlated in a linear relationship over a wide range of gene expression levels, despite differences in the measurement of their activity (FIAU accumulation in cells and GFP fluorescence) [59]. In the current study, the level of CD subunit expression measured in tumor tissue samples correlates highly with the radiotracer assay for TK subunit expression. These two data sets demonstrate the utility and validity of using fusion constructs for reporter gene imaging, where one of the subunits of the fusion construct is a reporter gene and can accurately “report” on the expression of the other subunit.
This combination of “reporter” and “therapeutic” subunits of a fusion gene is illustrated by the results reported here. We were able to generate parametric images of CD subunit activity in subcutaneous tumor xenografts based on the relationship that was established between the levels of [124I]FIAU accumulation (%dose/g) in the images and CD enzyme activity (CD U/g tumor) measured in tumor tissue sample homogenates. Knowing this relationship, parametric images of CD subunit activity could be generated as shown in Figure 3. In future studies using the CDglyTK fusion gene, the need of tissue sampling and direct measurements of CD subunit activity using the enzyme assays would not be necessary. Furthermore, it may be possible to rescale Figure 4 for human studies.
In summary, we demonstrated that imaging CDglyTK transgene expression is feasible using [124I]FIAU and PET. A highly consistent relationship between the enzymatic activities of the two subunits of the CDglyTK gene product, the CD-HSV1-TK fusion protein, was shown. Based on this relationship, the activity of the CD subunit can be derived from the FIAU radiotracer data in the images and parametric images of CD subunit activity generated. Parametric images of HSV1-TK and CD subunit activity in transduced tumor tissue can be used to define the sensitivity to either the 5FC, GCV, or 5FC/GCV combination therapy, Thus, PET imaging of CDglyTK expression could aid in the development, assessment, and optimization of clinical tumor gene therapy protocols by noninvasively defining the location, measuring the magnitude, and assessing the persistence of the therapeutic gene expression in target tissues.