Biomathematics, and 6UCLA-Jonsson Comprehensive Cancer Center, UCLA School of Medicine, ... ase (HSV1-tk), to particular cancer cells.1â6 In gene aug-.
Gene Therapy (2001) 8, 1072–1080 2001 Nature Publishing Group All rights reserved 0969-7128/01 $15.00 www.nature.com/gt
RESEARCH ARTICLE
Direct correlation between positron emission tomographic images of two reporter genes delivered by two distinct adenoviral vectors SS Yaghoubi1,2, L Wu4,6, Q Liang4, T Toyokuni1,2,3, JR Barrio1,3, M Namavari1,3, N Satyamurthy1,3, ME Phelps1,2,3,5, HR Herschman1,2,3,4,6 and SS Gambhir1,2,3,5,6 1
Department of Molecular and Medical Pharmacology, The Division of Nuclear Medicine, 2The Crump Institute for Biological Imaging, 3UCLA/DOE Laboratory of Structural Biology and Molecular Medicine, 4Molecular Biology Institute, 5Department of Biomathematics, and 6UCLA-Jonsson Comprehensive Cancer Center, UCLA School of Medicine, Los Angeles, CA, USA
Biodistribution, magnitude and duration of a therapeutic transgene’s expression may be assessed by linking it to the expression of a positron emission tomography (PET) reporter gene (PRG) and then imaging the PRG’s expression by a PET reporter probe (PRP) in living animals. We validate the simple approach of co-administering two distinct but otherwise identical adenoviruses, one expressing a therapeutic transgene and the other expressing the PRG, to track the therapeutic gene’s expression. Two PET reporter genes, a mutant herpes simplex virus type 1 thymidine kinase (HSV1-sr39tk) and dopamine-2 receptor (D2R), each regulated by the same cytomegalovirus (CMV) promoter, have been inserted into separate adenoviral vectors (Ad). We demonstrate
that cells co-infected with equivalent titers of Ad-CMV-HSV1sr39tk and Ad-CMV-D2R express both reporter genes with good correlation (r 2 = 0.93). Similarly, a high correlation (r 2 = 0.97) was observed between the expression of both PRGs in the livers of mice co-infected via tail-vein injection with equivalent titers of these two adenoviruses. Finally, microPET imaging of HSV1-sr39tk and D2R expression with 9-(4-[18F]fluoro-3hydroxymethylbutyl) guanine ([18F]FHBG) and 3-(2[18F]fluoroethyl)spiperone ([18F]FESP), utilizing several adenovirus-mediated delivery routes, illustrates the feasibility of evaluating relative levels of transgene expression in living animals, using this approach. Gene Therapy (2001) 8, 1072–1080.
Keywords: therapeutic transgene; positron emission tomography reporter genes; herpes simplex virus type 1 thymidine kinase; dopamine 2 receptor; adenoviral vectors
Introduction The ability to track the expression of a newly introduced transgene in living organisms may enhance progress in gene therapy. In particular, one would want to image the biodistribution, magnitude, and duration of a therapeutic gene’s expression once it is introduced in vivo. Confirming that the transgene is expressed primarily at its target site is particularly important when the therapeutic gene is a suicide gene, producing a cytotoxic protein. For example, many attempts have been made to deliver the suicide gene, herpes simplex virus type 1 thymidine kinase (HSV1-tk), to particular cancer cells.1–6 In gene augmentation therapy, where the purpose is to restore the magnitude of a normal gene’s expression to appropriate levels, it is essential to track the therapeutic gene’s expression level in the living subject.7–10 Measuring duration of expression of the therapeutic gene will also be of critical importance when using gene augmentation therapy in the clinic. Imaging the biodistribution of the
Correspondence: SS Gambhir, UCLA School of Medicine, 700 Westwood Plaza, B3-399A BRI, Los Angeles, CA 90095–1770, USA Received 17 November 2000; accepted 10 April 2001
therapeutic gene’s expression can also be used to predict possible side-effects of gene therapy. One possible approach to imaging gene expression in living animals, utilizes positron emission tomography (PET) reporter genes (PRG) and PET reporter probes (PRP).11–17 Current validated PRG/PRP systems measure either enzyme products of a reporter gene, which phosphorylates a radio-labeled substrate and traps it intracellularly, or receptor protein products of a reporter gene that binds a radio-labeled ligand.11–13 Examples of both types of PRG/PRP systems are the mutant HSV1-tk (HSV1-sr39tk) /9-(4-[ 18 F]fluoro-3-hydroxymethylbutyl) guanine ([18F]FHBG) and the dopamine-2 receptor (D2R)/3-(2-[18F]fluoroethyl)spiperone ([18F]FESP), both of which were recently validated.16,18–20 Through methods developed to establish a direct link between the expression of the therapeutic gene and the expression of a PRG, it should be possible to track biodistribution of a therapeutic gene’s expression and quantitatively determine the level of its expression, by imaging the PRG. Linking the expression of a therapeutic gene to a PRG can be done through a variety of different molecular constructs. One way is to clone both genes, side by side each other, downstream of a promoter, such that a single mRNA is transcribed from which a protein is translated that has the therapeutic and reporter proteins fused
Imaging dual gene expression with PET SS Yaghoubi et al
together.21,22 Another method is to construct a vector in which the therapeutic and the reporter genes are driven by a bidirectional promoter, transcripting separate mRNA from each gene which would then be translated into separate protein products.23–25 Yet another approach is to position both genes on separate locations of the same vector, downstream of an identical, but independent, promoter. Therefore both genes are activated together and two separate mRNAs and proteins are produced.26 A fourth novel approach is the incorporation of the therapeutic and the reporter gene in a bicistronic vector system in which a common promoter directs the transcription of a single mRNA, but an internal ribosomal entry site (IRES) separates the genes.27–30 In this approach a single mRNA will be produced from which the protein products of the therapeutic and reporter genes will each be separately translated. Recently, we described a bicistronic vector system in which a common cytomegalovirus (CMV) promoter directs transcription of a single mRNA containing coding regions of two reporter genes separated by the encephalomyocarditis virus IRES.28 The D2R reporter gene was placed upstream of IRES sequence and HSV1-sr39tk was placed downstream of the IRES (pCMV-D2R-IRES-HSV1sr39tk). Cells stably transfected with pCMV-D2R-IRESHSV1-sr39tk plasmid vector expressed both D2R and HSV1-sr39tk with high correlation. MicroPET imaging of tumors formed by subcutaneously implanting the same stably transfected tumor cells demonstrated that the expression of HSV1-sr39tk, as assessed by imaging with [18F]FHBG, could predict the expression of D2R, as assessed by [18F]FESP imaging. A similar approach has been employed by Tjuvajev et al,27 using single photon emission computerized tomography (SPECT) reporter probes for the HSV1-tk reporter gene. In the current study, we demonstrate that it is possible to macroscopically correlate the expression of two genes by regulating their transcription with the same promoter, but delivering them on separate adenoviral vectors. We make use of two adenoviruses, one carrying the HSV1sr39tk PRG and another carrying the D2R PRG. We demonstrate highly correlated expression of D2R and HSV1sr39tk in cultured cells co-infected with equal viral titers of Ad-CMV-HSV1-sr39tk and Ad-CMV-D2R and in the livers of mice co-injected intravenously with various equivalent titers of both adenoviruses. We confirm that it is possible to quantitatively image the expression of both PRGs with [18F]-FESP or [18F]-FHBG in various tissues of mice co-administered equivalent titers of AdCMV-HSV1-sr39tk and Ad-CMV-D2R; [18F]FHBG and [18F]FESP retention correlates in those tissues. Therefore, our study provides initial evidence that it should be possible to image biodistribution, magnitude and duration of a therapeutic transgene’s expression with a PRG, using this relatively simple approach.
Results Linearly increasing levels of D2R and TK expression are observed in C6 glioma cells infected with increasing titers of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk In order to confirm that the levels of expression of the PRGs increase in accordance with the viral titer used to infect cells, we incubated C6 cells with increasing titers
of Ad-CMV-HSV1-sr39tk (0 to 2.5 × 108 p.f.u.) or AdCMV-D2R (0 to 2.5 × 108 p.f.u.) for 48 h. Note that tk refers to the gene and TK refers to the enzyme. The control E1 deletion mutant (Ad5-dl434) was used to keep the total viral titer the cells were exposed to at 3.0 × 108 p.f.u. We observed a linear increase in TK enzyme activity or D2R receptor levels up to 2.5 × 108 p.f.u. (plus 0.5 × 108 p.f.u. Ad5-dl434) in C6 cells infected with Ad-CMVHSV1-sr39tk or Ad-CMV-D2R, respectively (Figure 1a
1073
Figure 1 (a) Correlation between TK expression in C6 cells and levels of Ad-CMV-HSV1-sr39tk infecting C6 cells. Twenty-four hours before infection 5 × 105 cells were plated in 100 mm petri dishes. The cells were incubated with 9 ml of media containing various titers of Ad-CMVHSV1-sr39tk for 48 h. Ad5-dl434 was used to keep viral titer constant at 3.0 × 108 p.f.u. (for example, the plate infected with 0.5 × 108 p.f.u. of Ad-CMV-D2R was also infected with 2.5 × 108 p.f.u. of Ad5-dl434). After 48 h the cells were trypsin-detached and harvested. The cells were lysed and their protein extracts were assayed for the level of TK enzyme activity. TK enzyme activity is expressed as percent conversion of [3H]-PCV converted to its phosphorylated form divided by 20 min of reaction time per g of total protein. The error bars indicate standard deviation based on triplicate determination of each sample. The correlation is r2 = 0.88 (y = 0.27x + 0.11). (b) Correlation between D2R expression in C6 cells and the amount of Ad-CMV-D2R infecting C6 cells. The cells were infected as above with varying titers of Ad-CMV-D2R for 48 h. Cells were harvested by scraping, and total protein was extracted after sonication. Total protein was assayed for binding of [3H]spiperone to D2R. D2R binding is expressed as the pmole of [3H]spiperone bound per mg total protein in cell lysates. The error bars indicate standard deviation based on triplicate determination of each sample. The correlation is r2 = 0.97 (y = 0.83x + 0.25). Gene Therapy
Imaging dual gene expression with PET SS Yaghoubi et al
1074
and b). The slope of the regression line in Figure 1a is 0.27 and in Figure 1b is 0.83. Within 0 to 2.5 × 108 p.f.u. of adenovirus infection in C6 cells, the levels of TK enzyme activity and D2R binding correlate well with the amount of adenoviral titer (r2 = 0.88 and 0.97, respectively). However, this fixed linear relationship between adenoviral titer and D2R binding or HSV1-sr39tk activity will not hold, if much higher amounts of adenovirus are used to infect C6 cells.
Cells co-infected with identical viral titers of Ad-CMVD2R and Ad-CMV-HSV1-sr39tk express both PRGs with good correlation In order to determine, in cell culture, whether the expression of a reporter gene can be used to estimate the expression of another transgene driven by the same promoter, we co-infected C6 or COS-1 cells with various equivalent titers (C6: 0–1.5 × 108 p.f.u.; COS1: 0–2.5 × 107 p.f.u.) of Ad-CMV-D2R and Ad-CMV-HSV1-sr39TK. Ad5-dl434 was used to keep the total viral titer C6 and COS-1 cells were exposed to constant at 3.0 × 108 and 5.0 × 107, respectively. After 48 h of incubation the coinfected cells were then lysed and assayed for TK enzyme activity and D2R binding. Figure 2a and b illustrate the activity of TK enzyme, stated as percent conversion of [3H]penciclovir to its monophosphorylated form per minute per g total protein in the lysate, versus pmoles of [3H]spiperone bound to D2R per mg protein. Regression analysis indicates high correlation (r2 = 0.78 and 0.93) between TK enzyme activity and D2R binding in C6 and COS1 cells, respectively. The slope of the regression lines are 2.11 and 5.71 in Figures 2a and b, respectively. C6 and COS-1 cells infected only with Ad-CMV-HSV1-sr39tk (2.0 × 108 p.f.u.) tested negative for D2R binding and the cells infected only with Ad-CMV-D2R (2.0 × 108 p.f.u.) tested negative for TK activity. Co-injection of equivalent titers of Ad-CMV-D2R and AdCMV-HSV1-sr39tk into the tail vein of mice leads to highly correlated expression of D2R and HSV1-sr39TK in their livers Eight Swiss Webster mice were co-injected various equivalent titers (0–1.5 × 109 p.f.u.) of Ad-CMV-D2R and AdCMV-HSV1-sr39tk into their tail veins (Ad5-dl434 was used to keep the total viral titer constant at 3.0 × 109 p.f.u.). Since adenoviruses administered intravenously primarily infect the liver, the livers of the dually adenovirus-infected mice were removed and assayed in vitro for TK enzyme activity and [3H]spiperone binding. Figure 3 illustrates TK enzyme activity versus D2R binding in the infected livers of the eight Swiss Webster mice. The expression of both PRGs is highly correlated (r2 = 0.97). The slope of the regression line is 2.2. MicroPET images illustrate correlated biodistribution and magnitude of [18F]FHBG and [18F]FESP activity in the livers of mice, co-injected with equivalent titers of AdCMV-HSV1-sr39tk and Ad-CMV-D2R intravenously To assess whether the expression of each of the PRGs can be tracked in living animals by imaging the expression of the other reporter gene, we co-injected various equivalent titers (0–1.5 × 109 p.f.u.) of Ad-CMV-D2R and Ad-CMVHSV1-sr39tk into the tail vein of 12 nude mice. Ad5-dl434 was used to keep the level of adenoviral exposure constant among all mice. After 2 days, we imaged the whole Gene Therapy
Figure 2 Correlation between TK enzyme activity and D2R binding in (a) C6 or (b) COS-1 cells co-infected with various equivalent amounts of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk. Twenty four hours after plating 5 × 105 cells in 100 mm dishes, the cells were co-infected with various equivalent amounts of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk in 9 ml fresh media. Ad5-dl434 was used to keep viral titer constant at 3.0 × 108 p.f.u. (for example, when COS-1 cells were co-infected with 1.0 × 107 p.f.u. of Ad-CMV-D2R and 1.0 × 107 p.f.u. of Ad-CMV-HSV1-sr39tk they were also infected with 3.0 × 107 p.f.u. of Ad5-dl434). After 48 h of co-infection, cells were harvested. Cells were lysed and total proteins were extracted. D2R binding assay determined pmole of [3H]spiperone bound per mg total protein and TK enzyme assay determined percent conversion of [3H]-PCV to its phosphorylated form in 20 min per g of total protein. The correlations are r2 = 0.78 (y = 2.11x + 0.62) and r2 = 0.93 (y = 5.71x + 1.55) in C6 and COS-1 cells, respectively. Error bars represent standard deviations based on triplicate determination of each sample.
body biodistribution of [18F]FHBG or [18F]FESP followed next day by microPET imaging of the same animals with the alternate reporter probe. Figure 4a shows the images of four of the mice scanned for [18F]FHBG and [18F]FESP activity. The images illustrate that as the [18F]FESP signal increases in the liver, the [18F]FHBG signal also increases. The percent injected doses per gram of tissues (%ID/g) of [18F]FHBG and [18F]FESP retention in the livers was determined. Figure 4b illustrates %ID/g retention of [18F]FHBG versus [18F]FESP in the livers of the same mice co-injected with various equivalent titers of Ad-CMVD2R and Ad-CMV-HSV1-sr39tk. The correlation between %ID/g retention of [18F]FHBG and [18F]FESP (r2 = 0.76) in the liver of 12 nude mice, indicate that we can estimate the expression of HSV1-sr39tk with [18F]FESP or D2R
Imaging dual gene expression with PET SS Yaghoubi et al
Figure 3 Correlation between TK enzyme activity and D2R binding in the liver of mice co-injected Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk in the tail vein. Adult Swiss Webster mice were tail-vein injected with 200 l of 0.9% NaCl containing equal viral titers of both Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk, increased to a constant level of 3.0 × 109 p.f.u. with Ad5-dl434. Forty eight hours later the mice were killed and small pieces of their livers were homogenized and prepared for TK enzyme activity and D2R binding assays. The graph illustrates correlation between the expression of D2R and HSV1-sr39TK proteins as described by percent conversion of [3H]-PCV to its phosphorylated form/reaction time/g total protein versus pmole [3H]spiperone bound/mg total protein (r2 = 0.97; y = 2.2x + 0.03). The error bars indicate standard deviation of triplicate determination of samples from liver homogenates of each mouse. Note that two of the data points are very close to each other and are visible as a single data point near zero.
with [18F]FHBG. Each data point is an average %ID/g of three ROI measurements over the liver in the same image.
Retention of [18F]FHBG and [18F]FESP in the muscle and tumor implants of nude mice infected with Ad-CMVHSV1-sr39tk and Ad-CMV-D2R confirms that imaging transgene expression may be extended to extra-hepatic tissues To examine whether our dual PRG imaging method is applicable in extrahepatic tissues, we co-injected equivalent titers of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk into muscles of nude mice, and into tumor xenografts produced subcutaneously in nude mice, and then imaged these animals with [18F]FHBG and [18F]FESP. The tumorcarrying mice were also imaged with 2-deoxy-2[18F]fluoro-D-glucose ([18F]FDG), to make sure the tumors consisted of viable cancer cells.31 1.0 × 109 p.f.u. of both Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk were injected into the right hind leg (final concentration of both viruses was 1.0 × 107 p.f.u./l in a total volume of 100 l), and 0.5 × 109 p.f.u. of both Ad-CMV-D2R and Ad-CMV-HSV1sr39tk was injected into the left hind leg (final concentration of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk was 0.5 × 107 p.f.u./l adjusted by adding 50 l of Ad5-dl434 to get a total volume of 100 l) of a nude mouse, 2 days before the start of imaging. Only Ad5-dl434 control was injected into the right paw of the nude mouse. A nude mouse was first imaged with [18F]FESP, 2 days after virus injection, then the next day by [18F]FHBG. The highest
activity of [18F]FHBG (0.89 ± 0.33 %ID/g) was observed in the right hind leg and corresponded with the highest activity of [18F]FESP (1.76 ± 0.83 %ID/g) in the same leg. Lower activity of [18F]FHBG (0.19 ± 0.12 %ID/g) in the left hind leg corresponded with a lower activity of [18F]FESP (0.92 ± 0.19 %ID/g) in that leg. However, background activity of [18F]FHBG (0.02 ± 0.01 %ID/g) was observed in the right paw, which corresponded with the background activity of [18F]FESP (0.58 + 0.16 %ID/g), in the right paw. Activity of [18F]FHBG and [18F]FESP were observed in the livers of most of the mice injected intramuscularly, due to leakage of adenoviruses into blood vessels. The nude mouse shown in Figure 5 had one tumor xenograft over his neck (tumor 1) and one tumor xenograft over his buttocks (tumor 2). Tumor 1 was coinjected 1.0 × 109 p.f.u. of both Ad-CMV-D2R and AdCMV-HSV1-sr39tk on 3 consecutive days, before being imaged with [18F]FESP. Tumor 2 was injected with only the control virus Ad5-dl434. The day after imaging with [18F]FESP, the mouse was imaged with [18F]FDG. Two days after imaging with [18F]-FESP the same mouse was imaged with [18F]FHBG. As shown in Figure 5, the [18F]FESP signal (3.36 ± 0.27 %ID/g) from tumor 1 corresponds with the [18F]FHBG signal (6.30 ± 0.27 %ID/g) from the same tumor. On the other hand both [18F]FESP (0.05 ± 0.01 %ID/g) and [18F]FHBG (0.38 ± 0.06 %ID/g) signals from tumor 2 are at the background level. However, [18F]FDG activity was detected from both tumor 1 (1.92 ± 0.14 %ID/g) and tumor 2 (1.59 ± 0.16 %ID/g). Activity of [18F]FHBG and [18F]FESP were observed in the livers of some of the mice injected intratumorally, due to leakage of adenoviruses into blood vessels.
1075
Biodistribution and magnitude of D2R and HSV1-sr39TK expression is correlated at different time points To assess whether the expression of both PET reporter transgenes remains correlated in living mice over time, we co-injected equivalent titers of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk into the tail veins of both Swiss Webster mice (immunocompetent) and nude mice (immunodeficient). We followed expression of both PRGs, by [18F]FHBG and [18F]FESP scanning, at different days after the co-injection of the adenoviruses, using the microPET. Figure 6a illustrates the activity of [18F]FESP (upper panel) and [18F]-FHBG (lower panel) at various times (extending up to a month) after the intravenous coinjection of 2.0 × 109 p.f.u. of Ad-CMV-D2R and Ad-CMVHSV1-sr39tk in a Swiss Webster mouse. The activity of both reporter probes declines linearly with a correlation of r2 = 0.997 in the liver of the mouse, over time. These results were confirmed by imaging a second Swiss Webster mouse (data not shown). Figure 6b is a demonstration of the same experiment in a nude mouse. The nude mouse was intravenously co-injected with 0.1 × 109 p.f.u. of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk and imaged at three different time-points extending to a month after the injection of the adenoviruses. The activity of both probes in the liver increased from the first to the second week after adenovirus injection and then remained relatively constant 2 weeks after that. The correlation between activity of [18F]FHBG and [18F]FESP was good (r2 = 0.81). Gene Therapy
Imaging dual gene expression with PET SS Yaghoubi et al
1076
a
Figure 4 (a) MicroPET whole body [18F]FESP and [18F]FHBG images of nude mice co-injected Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk intravenously. Equal viral titers of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk were dissolved in 200 l 0.9% NaCl and injected into the tail vein of nude mice. After 3 days, the mice were whole body scanned for 28 min (4 min per each of seven beds) with MicroPET for signals from either [18F]FHBG or [18F]FESP, injected through the tail vein. The next day, the same mice were scanned identically with the alternate PRP. Images illustrate degree of accumulation and biodistribution of [18F]FHBG and [18F]FESP in individual nude mice co-injected with different, but equivalent titers of Ad-CMVHSV1-sr39tk and Ad-CMV-D2R. From left to right, the first mouse received 0 p.f.u., the second mouse received 0.4 × 108 p.f.u., the third mouse received 0.5 × 108 p.f.u. and the fourth mouse received 3.6 × 108 p.f.u. The upper panel illustrates biodistribution of [18F]FESP and the lower panel that of [18F]FHBG. Images are the average of several whole body slices containing the liver. The images are scaled based on the injected dose. Images shown are based on MAP reconstruction of the microPET data. (b) Quantitative determination of correlation between [18F]FHBG and [18F]FESP activity in the livers of nude mice injected with equivalent doses of Ad-CMV-HSV1-sr39tk and Ad-CMV-D2R. Twelve nude mice (four of whom were shown in panel a) were co-injected intravenously with various equivalent titers of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk. The mice were imaged with microPET for the detection of [18F]FHBG and [18F]FESP (which had been injected through tail veins), on 2 consecutive days. Whole body images were reconstructed by filtered back projection. The whole body slices containing the liver were averaged. Three different regions of interest (ROI) were drawn over the livers and average counts/second/pixel in each ROI was determined. The graph illustrates %ID/g retention of [18F]FHBG versus %ID/g retention of [18F]FESP in the livers of the nude mice (r2 = 0.76; y = 1.49x + 1.06). The error bars are standard deviations of the three average %ID/g.
Discussion In this study we have demonstrated that the expression of the two reporter transgenes, HSV1-sr39tk and D2R, driven by the same constitutively active CMV promoter is well correlated at the multi-cell (macroscopic) and tissue level, even though both reporter genes are delivered utilizing separate adenoviral vectors. We observed good correlation between the activity of TK enzyme and D2R binding in two different cell types, that had been coinfected with varying but equivalent titers of Ad-CMVHSV1-sr39tk and Ad-CMV-D2R, in cell culture. Furthermore, we observed excellent correlation between the penciclovir-phosphorylating activity of TK enzyme and spiperone binding of D2R in the livers of mice following coinjection of equivalent titers of Ad-CMV-HSV1-sr39tk and Ad-CMV-D2R. HSV1-sr39TK enzyme more effectively utilizes acycloguanosines such as ganciclovir ([3H]GCV, and [18F]GCV) and penciclovir ([3H]PCV, [18F]PCV, and [18F]FHBG) as substrates than the wildtype HSV1-TK; thus exhibits improved sensitivity for imaging PRG expression with PET.18 Our approach is likely to be applicable if wild-type HSV1-tk PRG is used, however at relatively low levels of HSV1-tk expression, Gene Therapy
there may be insufficient signal to detect over background levels. The liver, muscle and tumor images in our study have shown that once Ad-CMV-D2R and Ad-CMV-HSV1sr39tk are co-injected together, [18F]FESP can not only track the biodistribution of D2R, but it can also track and predict the bio-distribution of HSV1-sr39TK. Similarly, [18F]FHBG can not only track the bio-distribution of HSV1-sr39TK, but also D2R. Therefore, it should be possible to infer the biodistribution of a therapeutic gene’s expression, delivered by an adenoviral vector and regulated by the CMV promoter, in a living animal by coinjecting it with either Ad-CMV-D2R or Ad-CMV-HSV1sr39tk and imaging with [18F]FESP or [18F]FHBG, respectively. However, this approach may also be effective, if a promoter other than the CMV promoter regulates the expression of the therapeutic gene and the PRG. It may also be possible to track the biodistribution of a therapeutic gene, delivered by other types of gene delivery vectors by co-administering it with an identical vector carrying the PRG. Our study provides initial evidence with a replication-deficient adenovirus vector and the CMV promoter and must be validated for use with other promoters or DNA delivery vectors.
Imaging dual gene expression with PET SS Yaghoubi et al
1077
a
Figure 5 [18F]-FESP, [18F]-FDG and [18F]-FHBG images of C6 tumor xenograft co-injected with equivalent titers of Ad-CMV-HSV1-sr39tk and Ad-CMV-D2R. The nude mouse shown in Figure 5 had one tumor xenograft over its neck and one tumor xenograft over its buttocks. Tumors were grown subcutaneously after implanting 1 × 107 C6 cells. Adenoviruses were injected 10 days later. The neck tumor (tumor 1) was co-injected with 1.0 × 109 p.f.u. of both Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk on 3 consecutive days, before being imaged with [18F]FESP. The buttock tumor (tumor 2) was injected only with the control virus Ad5-dl434. The day after imaging with [18F]FESP, the mouse was imaged with [18F]FDG followed the next day by imaging with [18F]FHBG. The PET (whole-body) image is the sum of all planes of the [18F]FDG scanned image. The [18F]FDG, [18F]FESP and [18F]FHBG PET (section) images are averages of the planes that contain the tumors. The PET (section) images have been scaled to the injected dose. This mouse is representative of a larger data set.
Perhaps the greatest advantage of our approach is the convenience it will yield to animal or human therapeutic transgene pharmacokinetic experiments. This approach may eliminate the need for making a new construct every time an investigator needs to track the expression of the therapeutic gene in vivo. In fact many investigators have already delivered therapeutic genes regulated by the widely used CMV promoter. The idea behind this current approach is that host responses to a gene therapy vector will be the same whether it is carrying the therapeutic gene or if it is carrying a PRG. Therefore, either vector has an equal chance of uptake by the types of cells capable of infection by that specific type of vector. Furthermore, host responses that somehow eliminate the vector carrying the therapeutic gene will have an equal chance of eliminating the identical vector carrying the PRG. Therefore, one would speculate that a lack of correlation between the expression of a therapeutic gene and a PRG delivered by this approach would have to be due to genespecific regulation related to the nature of the expressed protein. Once validated in cell culture and in animal models that the PRG and the therapeutic gene which are driven by the same promoter are identically regulated, independent reporter gene vectors may be delivered simultaneously with the therapeutic gene vectors to track the biodistribution, duration and the level of expression of the therapeutic gene in living animals and in patients. The fact that the identical promoters regulating the expression of the therapeutic gene and the PRG are on separate vectors may affect correlation between the genes. Cis-acting mechanisms of transcription regulation, due to positive and negative feedback from either the therapeutic gene or the PRG may reduce the correlation between their expression. This may not be a problem with the IRES, fusion protein and bidirectional promoter
b
Figure 6 [18F]FESP and [18F]FHBG coronal images of (a) a Swiss Webster and (b) a nude mouse at three different time-points after co-injecting equivalent titers of Ad-CMV-HSV1-sr39tk and Ad-CMV-D2R intravenously. (a) 2.0 × 109 and (b) 0.1 × 109 p.f.u. of both Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk were co-injected into the tail vein of a Swiss Webster mouse and a nude mouse, respectively. The mice were then scanned for [18F]FESP and [18F]FHBG activity. For the [18F]FESP images times 1, 2 and 3 correspond to 3, 10 and 24 days after the injection of the adenoviruses, respectively. For the [18F]FHBG images times 1, 2 and 3 correspond to 5, 12 and 26 days after the injection of the adenoviruses, respectively. The images are scaled for injected dose. These mice are representatives of a larger data set.
approaches for delivery of therapeutic and reporter genes, because any affect on the promoter will affect the transcription of both the therapeutic gene and the PET reporter gene. However, at the translational and posttranslational level all the approaches, except the fusion protein approach, may encounter loss of correlation between the protein products of the therapeutic gene and the PRG. In the bicistronic IRES construct, the expression of the gene downstream of the IRES sequence is attenuated, which might affect sensitivity of the reporter gene’s Gene Therapy
Imaging dual gene expression with PET SS Yaghoubi et al
1078
detection or the efficacy of the therapeutic gene. Recently, IRES sequences were identified which may help to solve this problem.32 In the fusion protein approach, the final quaternary folding of the fusion protein might affect the function of the therapeutic gene or influence the ability of the reporter probes to interact with reporter gene product. Whichever approach is used for imaging a new therapeutic gene, the correlation between expression of the therapeutic gene and PRG product should be examined in cell culture and animal models. Another argument against delivering the therapeutic gene and the PRG on separate vectors is that not all cells may be equally infected with the PRG vector and the therapeutic gene vector. In contrast, in the other approaches both genes are constructed on the same vector; thus at the single cell level equal numbers of both genes are delivered. However, in PET imaging we cannot image single cells; we are concerned with the macroscopic scale of multicell substructure and tissues. Once administered into the living animal, identical vectors carrying the therapeutic and PET reporter genes have an equal statistical chance of infecting each cell. Therefore, a reasonable correlation between infection of both vectors can be expected at the macroscopic level. Our results provide evidence for this hypothesis. Another limitation of delivering therapeutic and reporter genes on separate adenoviruses is the concomitant increase of viral burden and a greater chance of immunological reactions. In the studies presented here we co-administered a control replication-deficient adenovirus to be certain that all animals received the identical viral burden when lower doses of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk were administered. Furthermore, with the use of less immunogenic adenoviral vectors the increased viral burden may not be a problem. For the purpose of quantifying the expression levels of a therapeutic gene from quantitative PET images of a linked PRG an in vivo fixed linear relationship should be determined. However, it should be noted that both in vitro and in vivo, there is a specific range of PRG and therapeutic gene expression levels at which a reasonable correlation and a fixed linear relationship exists. At high viral titers (0.5–1.5 × 109 p.f.u.) we reached saturation of reporter probe activity in the livers of nude mice, as observed by microPET liver images. Furthermore, the level of gene expression is cell type dependent. For example, the higher range of multiplicity of infections (MOI) of the adenoviruses used to infect the C6 cells (0– 300 MOI) resulted in lower PRG expression levels than the lower range (0–50 MOI) used to infect COS-1 cells (refer to Figure 2). Therefore, when therapeutic genes are co-administered with PRG, in practice, some preliminary studies will be required to determine the appropriate range where a fixed linear relationship exists between the expression of both genes. In the current work we have used %ID/g as a quantitative measure of reporter gene expression. This is a relatively crude measure as it does not take advantage of kinetic information for a given reporter probe to separate out processes such as flow, uptake, trapping, etc. We are currently developing pharmacokinetic models33 and should be able to improve correlations by more accurate quantitation of reporter gene expression. A good correlation was also observed at different timepoints of imaging, up to a month between the activity of
Gene Therapy
[18F]FHBG and [18F]FESP in the liver of mice intravenously co-injected with equal titers of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk (refer to Figure 6a and b). In fact, if we quantify the tracer signals in the liver of the nude mouse shown in Figure 6b, those data fit into the fixed linear relationship shown in Figure 4b. Therefore, these data suggest that in practice the correlation between the expression of the therapeutic gene and the PRG may hold up for a substantial time following coadministration of the two viruses. In conclusion, it is possible to track the biodistribution, magnitude and duration of a therapeutic gene delivered in an adenovirus vector by simply co-administering an otherwise identical adenovirus vector expressing a PET reporter gene.
Materials and methods Radiolabeled compounds [8-3H]-penciclovir ([8-3H]-PCV) (13.2 Ci/mmole) was purchased from Moravek Biochemicals, Inc. (LaBrea, CA, USA). [3H]spiperone (16.5 Ci/mmole) was purchased from NEN (Boston, MA, USA). 9-(4-[18F]-fluoro-3hydroxymethylbutyl)guanine ([18F]FHBG) and 3-(2⬘-[18F]fluoroethyl)spiperone ([18F]FESP) were synthesized as described previously at specific activities of between 1000 and 2000 Ci/mmol.19,34 Note that [18F]FHBG is the sidechain fluorine-18 radiolabeled analog of penciclovir. 2Deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) was synthesized by the method described by Hamacher et al35 at specific activities of approximately 5000 Ci/mmol. Radioactivity determinations All tritium analysis was performed with a Beckman LS6500 Liquid Scintillation Counter and Cytoscint (ICN, Costa Mesa, CA, USA) as scintillation fluid. Disintegrations per minute (d.p.m.) were obtained by correcting for background activity and efficiency (67%) based on calibrated standards (Beckman, Inc., Fullerton, CA, USA). Adenoviral vectors Dopamine-2 receptor adenovirus (Ad-D2R) was constructed by inserting the cloned rat D2R coding sequence behind CMV promoter into the E1 region of a replication defective type 5 adenovirus.16 Construction of HSV1-tk adenovirus (Ad-CMV-HSV1-tk) was performed as described previously.14 However, for the experiments reported here, the mutant HSV1-sr39tk gene was inserted between the CMV immediate–early promoter and the SV40 splice and poly A site.36 The control virus, Ad5dl434, is an E1 deletion mutant with viral sequences between 2.6 and 8.7 map units deleted.37 Cell lines C6 rat glioma cells were provided by Dr M Black (Chiroscience R&D Inc., Bothell, WA, USA). COS-1 cells had been obtained from ATCC. C6 cells were grown in Deficient D-MEM High Glucose (Life Technologies, Gaithersburg, MD, USA) supplemented with 10% FBS, 1% penicillin-streptomycin, and 1% l-glutamine. COS-1 cells were grown in D-MEM with 4 mm l-glutamine, 1.5 g/l sodium bicarbonate, 4.5 g/l glucose and 1.0 mm sodium pyruvate (Life Technologies) supplemented with 10% FBS and 1% penicillin-streptomycin.
Imaging dual gene expression with PET SS Yaghoubi et al
Adenoviral co-infection of cultured cells Twenty-four hours before infection, C6 or COS-1 cells were plated in 100-mm Petri dishes at a count of 5 × 105 cells per dish. To obtain the linear range at which the expression of HSV1-sr39tk and D2R increases linearly with increasing titers of the adenoviral vectors, we infected C6 cells with each of the adenoviruses (0 to 3 × 108 p.f.u.) separately in 9 ml of media. To examine correlation between HSV1-sr39tk and D2R in cell culture, C6 or COS-1 cells were co-infected with various equivalent titers (C6: 0–1.5 × 108 p.f.u.; COS1: 0–2.5 × 107 p.f.u.) of Ad-CMV-D2R and Ad-CMV-HSV1-sr39tk. The C6 cells in each plate were exposed to a total viral burden of 3 × 108 p.f.u. and COS-1 cells in each plate were exposed to a total viral burden of 5.0 × 107 p.f.u., by adding enough Ad5-dl434 virus. The total time of exposure to adenovirus, before the cells were assayed for HSV1-sr39tk and D2R expression was 48 h. Adenoviral co-administration into mice For in vitro analysis of gene expression in mouse livers, equivalent titers of Ad-CMV-D2R and Ad-CMV-HSV1sr39tk (between 0 to 1.5 × 109 p.f.u.) were co-injected into the tail veins of Swiss Webster mice, 48 h before killing them and extracting their livers for TK enzyme and D2R binding assays. Ad5-dl434 was added to keep the total viral titer at 3.0 × 109 p.f.u. Tail vein injection of adenovirus causes a predominant infection of the mouse liver and gene expression peaks within 48 h.38–40 For wholebody imaging of the PRGs with [18F]FESP and [18F]FHBG, various equivalent titers of Ad-CMV-D2R and Ad-CMVHSV1-sr39tk (between 0 to 1.5 × 109 p.f.u.) were coinjected into the tail veins of the immunodeficient nude mice. Ad5-dl434 was added to the adenoviral mix, if needed, to bring the total viral titer to a constant level. For in vivo imaging of the PRGs in mouse muscle and implanted tumors, equivalent titers of both viruses were injected directly into the muscles or the tumors. D2R binding assay Cells stored free of media at −20°C were harvested by scraping in 6 ml of TE buffer (10 mm Tris-HCl; 1 mm EDTA; pH = 7.5). The cells were then pelleted and resuspended in 200 l TE buffer. Cell lysates were then prepared by sonication. Proteins were then separated from chunks by centrifugation at 400 g. Cell lysates were then diluted at 25 g total protein per 200 l assay solution (5 mm Tris-HCl pH = 7.4; 150 mm NaCl; 0.025% ascorbic acid: 0.001% bovine serum albumin). For in vitro assay of D2R expression in the liver, small pieces of liver were homogenized in a buffer consisting of 25 mm Tris (pH = 7.5), 6 mm MgCl2, 2 mm EDTA, 0.25 m sucrose, 300 m PMSF and 1 mm benzamidine on the day of extraction. 500 l of the homogenate was then sonicated for 10 s and the homogenate was spun down at 400 g. The supernatants from liver homogenates were diluted at 150 g total protein per 200 l assay solution. Samples were incubated with 3 nm [3H]spiperone in the presence or absence of 10 m (+)-butaclamol (Research Biochemical International, Natick, MA, USA). After incubation on ice for 30 min, samples were filtered through 24 mm GF/F glass microfiber filters (Whatman, Clifton, NJ, USA). The filters were washed three times with 5 ml each of cold (4°C) wash solution (50 mm Tris-HCl pH 7.4, 150 mm NaCl) and placed in scintillation vials with 5 ml Cytoscint
scintillation fluid (ICN Pharmaceuticals, Inc., Costa Mesa, CA, USA) overnight, before radioactivity determination.
1079
TK enzyme assay Cells were harvested by trypsinization and pelleted. The cells were lysed by dissolving the pellet in TK lysis buffer (10 mm Tris-HCl pH = 7.5; 3 mm 2-mercaptoethanol; 0.5% Igepal CA-630; 25 mm NaF) and twice freeze–thawing. Liver pieces were homogenized on the day of extraction in 4 ml of TK lysis buffer. Proteins were separated by centrifuging in a microcentrifuge at 14 000 r.p.m., 4°C, 20 min. Protein concentrations in the supernatant were determined by the BioRad Protein Microassay procedure. Cell (1 g total protein) or liver (2 g total protein) lysates were incubated in a 37°C water bath for 20 min with [3H]-PCV (53 pmoles) in TK assay buffer (0.25 m NaPO2 pH 6.0; 25 mm ATP pH 7.0; 25 mm Mg-acetate) to phosphorylate [3H]-PCV. [3H]-PCV was separated from phosphorylated [3H]-PCV using a DE81 filter (Whatman).41 Results are reported as percent conversion of total [3H]PCV available in reaction mixture per min per g protein. MicroPET imaging Mice were injected via the tail vein with [18F]FHBG (180– 250 Ci), [18F]FESP (200–350 Ci) or [18F]FDG (only for nude mice carrying C6 tumors). The mice were held for 1 h after the injection of [18F]FHBG, 3 h after the injection of [18F]FESP and 1 h after the injection of [18F]FDG (anesthetized during the waiting period) before being scanned. Immediately before scanning the mice were anesthetized with ketamine/xylazine and imaged for 28 min using a microPET scanner with a 23 mm3 volumetric resolution. The long axis of the mouse was parallel to the long axis of the scanner and 7-bed positions with 4 min per bed were used. Images were reconstructed by using filtered-back projection and an iterative three-dimensional reconstruction technique with an isotropic image resolution of 1.8 mm and a volumetric resolution of approximately 8 mm3.42 Quantitation of fluorine-18 retention in the livers, muscles and implanted tumors in nude mice Regions of interest (ROI) were drawn over left liver lobe, muscle, or tumor on decay corrected whole body coronal images. The counts/pixel/min obtained from the ROI were converted to counts/ml/min by using a calibration constant obtained from scanning a cylinder phantom in the microPET scanner. The ROI counts/ml/min were converted to counts/g/min, assuming a tissue density of 1 g/ml, and divided by the injected dose to obtain an image ROI derived percent injected dose of radionuclide per gram of tissue.
Acknowledgements We thank the members of the UCLA Gene Imaging Consortium, K Nguyen, E Bauer and Y Yu for various technical assistance, R Sumida, W Ladno, J Edwards, DJ Liu and V Dominguez for assistance with microPET imaging, M Ho for the synthesis of [18F]FESP and [18F]FHBG and the UCLA cyclotron crew for outstanding support. We are grateful for J Matherly’s assistance with adenovirus replication. We thank A Green for helpful discussions. Grants: this work was partially supported by funding Gene Therapy
Imaging dual gene expression with PET SS Yaghoubi et al
1080
from NIH RO1 CA82214–01, DOE contract DE-FC03– 87ER60615, University of California Biotechnology Grant, Dana Foundation, and the UCLA-Jonsson Comprehensive Cancer Center.
22
References
23
1 Osaki T et al. Gene therapy for carcinoembryonic antigen-producing human lung cancer cells by cell type-specific expression of herpes simplex virus thymidine kinase gene. Cancer Res 1994; 54: 5258–5261. 2 Kaneko S et al. Adenovirus-mediated gene therapy of hepatocellular carcinoma using cancer-specific gene expression. Cancer Res 1995; 55: 5283–5287. 3 Vandier D et al. Selective killing of glioma cell lines using an astrocyte-specific expression of the herpes simplex virus-thymidine kinase gene. Cancer Res 1998; 58: 4577–4580. 4 Vile RG, Hart IR. Use of tissue-specific expression of the herpes simplex virus thymidine kinase gene to inhibit growth of established murine melanomas following direct intratumoral injection of DNA. Cancer Res 1993; 53: 3860–3864. 5 DiMaio JM et al. Directed enzyme pro-drug gene therapy for pancreatic cancer in vivo. Surgery 1994; 116: 205–213. 6 Smith MJ et al. Surfactant protein A-directed toxin gene kills lung cancer cells in vitro. Hum Gene Ther 1994; 5: 29–35. 7 Cannizzo SJ et al. Augmentation of blood platelet levels by intratracheal administration of an adenovirus vector encoding human thrombopoietin cDNA. Nat Biotechnol 1997; 15: 570–573. 8 Kay MA et al. In vivo hepatic gene therapy: complete albeit transient correction of factor IX deficiency in hemophilia B dogs. Proc Natl Acad Sci USA 1994; 91: 2353–2357. 9 Factor P et al. Augmentation of lung liquid clearance via adenovirus-mediated transfer of a Na,K-ATPase b1 subunit gene. J Clin Invest 1998; 102: 1421–1430. 10 Cable DG et al. Recombinant endothelial nitric oxide synthasetransduced human saphenous veins: gene therapy to augment nitric oxide production in bypass conduits. Circulation 1997; 96 (Suppl. II): II-173–II-178. 11 Gambhir SS et al. Assays for noninvasive imaging of reporter gene expression. Nucl Med Biol 1999; 26: 481–490. 12 Gambhir SS et al. Imaging gene expression: principles and assays. J Nucl Cardiol 1999; 6: 219–233. 13 Herschman HR et al. Seeing is believing: non-invasive, quantitative and repetitive imaging of reporter gene expression in living animals, using positron emission tomography. J Neurosci Res 2000; 59: 699–705. 14 Gambhir SS et al. Imaging of adenoviral directed herpes simplex virus type 1 thymidine kinase gene expression in mice with ganciclovir. J Nucl Med 1998; 39: 2003–2011. 15 Gambhir SS et al. Imaging adenoviral-directed reporter gene expression in living animals with positron emission tomography. Proc Natl Acad Sci USA 1999; 96: 2333–2338. 16 MacLaren DC et al. Repetitive, non-invasive imaging of the dopamine D2 receptor as a reporter gene in living animals. Gene Therapy 1999; 6: 785–791. 17 Tjuvajev JG et al. Imaging herpes virus thymidine kinase gene transfer and expression by positron emission tomography. Cancer Res 1998; 58: 4333–4341. 18 Gambhir SS et al. A mutant herpes simplex virus type 1 thymidine kinase reporter gene shows improved sensitivity for imaging reporter gene expression with positron emission tomography. Proc Natl Acad Sci USA 2000; 97: 2785–2790. 19 Alauddin MM, Conti PS. Synthesis and preliminary evaluation of 9-(4-[18F]-fluoro-3-hydroxymethylbutyl)guanine ([18F]FHBG): a new potential imaging agent for viral infection and gene therapy using PET. Nucl Med Biol 1998; 25: 175–180. 20 Gambhir SS et al. Imaging transgene expression with radionuclide imaging technologies. Neoplasia 2000; 2: 118–138. 21 Loimas S, Wahlfors J, Janne J. Herpes simplex virus thymidine kinase-green fluorescent protein fusion gene: new tool for gene
Gene Therapy
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41 42
transfer studies and gene therapy. Biotechniques 1998; 24: 614– 618. Rainov NG et al. A chimeric fusion protein of cytochrome CYP4B1 and green fluorescent protein for detection of pro-drug activating gene delivery and for gene therapy in malignant glioma. Adv Exp Med Biol 1998; 451: 393–403. Bell PJL et al. A two-reporter gene system for the analysis of bi-directional transcription from the divergent MAL6T-MAL6S promoter in Saccharomyces cerevisiae. Curr Genet 1995; 28: 441– 446. Baron U et al. Co-regulation of two gene activities by tetracycline via a bidirectional promoter. Nucleic Acids Res 1995; 23: 3605– 3606. Sun X et al. Quantitation of gene induction by PET imaging of a bi-directional tetracycline-inducible reporter gene system in living animals. J Nucl Med 2000; 41: 38P (Abstr. 147). Bramson J et al. Construction of a double recombinant adenovirus vector expressing a heterodimeric cytokine: in vitro and in vivo production of biologically active interleukin-12. Hum Gene Ther 1996; 7: 333–342. Tjuvajev JG et al. A general approach to the non-invasive imaging of transgenes using cis-linked herpes simplex virus thymidine kinase. Neoplasia 1999; 1: 315–320. Yu Y et al. Quantification of target gene expression by imaging reporter gene expression in living animals. Nat Med 2000; 6: 933–937. Wiznerowicz M et al. Development of a double-copy bicistronic retroviral vector for human gene therapy. Adv Exp Med Biol 1998; 451: 441–447. Pizzato M et al. Production and characterization of a bicistronic Moloney-based retroviral vector expressing human interleukin 2 and herpes simplex virus thymidine kinase for gene therapy of cancer. Gene Therapy 1998; 5: 1003–1007. Chung JK et al. The effect of tumor size on F-18-labeled fluorodeoxyglucose and fluoroerythronitroimidazole uptake in a murine sarcoma model. Ann Nucl Med 1999; 13: 303–308. Chappell SA, Edelman GM, Mauro VP. A 9-nt segment of a cellular mRNA can function as an internal ribosome entry site (IRES) and when present in linked multiple copies greatly enhances IRES activity. Proc Natl Acad Sci USA 2000; 97: 1536– 1541. Green LA et al. Tracer kinetic modeling of FHBG in mice imaged with MicroPETfor quantitation of reporter gene expression. J Nucl Med 2000; 41: 58P (Abstr. 228). Satyamurthy N et al. 3-(2⬘-[18F]fluoroethyl)Spiperone, a potent dopamine antagonist: synthesis, structural analysis and in vivo utilization in human. Appl Radiat Isotopes 1990; 41: 113–129. Hamacher K, Coenen H, Stocklin G. Efficient stereospecific synthesis of no-carrier-added 2[18F]-fluoro-2-deoxy-D-glucose using amino polyether supported nucleophilic substitution. J Nucl Med 1986; 27: 235–238. Black ME et al. Creation of drug-specific herpes simplex virus type 1 thymidine kinase mutants for gene therapy. Proc Natl Acad Sci USA 1996; 93: 3525–3529. Grodzicker T, Klessig D. Expression of unselected adenovirus genes in human cells co-transformed with the HSV-1 tk gene and adenovirus 2 DNA. Cell 1980; 21: 453–463. Stratford-Perricaudet LD et al. Evaluation of the transfer and expression in mice of an enzyme-encoding gene using a human adenovirus vector. Hum Gene Ther 1990; 1: 241–256. Herz J, Gerard RD. Adenovirus-mediated transfer of low density lipoprotein receptor gene acutely accelerates cholesterol in normal mice. Proc Natl Acad Sci USA 1993; 90: 2812–2816. Srinivasan A et al. A PET reporter gene (PRG)/PET reporter probe (PRP) technology for repeatedly imaging gene expression in living animals. J Nucl Med 1996; 37: 107P. Hruby D, Ball A. Cell free synthesis of enzymatically active vaccinia virus thymidine kinase. Virology 1981; 113: 594–601. Qi J et al. Fully 3D Bayesian image reconstruction for the ECAT EXACT HR+. IEEE Trans Nuclear Sci 1998; 45: 1096–1103.