Utility of TK/GCV in the context of highly effective oncolysis mediated by a serotype 3 receptor targeted oncolytic adenovirus

Abstract

Arming oncolytic adenoviruses with therapeutic transgenes and enhancing transduction of tumor cells are useful strategies for eradication of advanced tumor masses. Herpes simplex virus thymidine kinase (TK) together with ganciclovir (GCV) has been promising when coupled with viruses featuring low oncolytic potential, but their utility is unknown in the context of highly effective infectivity-enhanced viruses. We constructed Ad5/3-Δ24-TK-GFP, a serotype 3 receptor-targeted, Rb/p16 pathway-selective oncolytic adenovirus, where a fusion gene encoding TK and green fluorescent protein (GFP) was inserted into 6.7K/gp19K-deleted E3 region. Ad5/3-Δ24-TK-GFP killed ovarian cancer cells effectively, which correlated with GFP expression. Delivery of GCV immediately after infection abrogated viral replication, which might have utility as a safety switch. Due to the bystander effect, killing of some cell lines in vitro was enhanced by GCV regardless of timing. In murine models of metastatic ovarian cancer, Ad5/3-Δ24-TK-GFP improved antitumor efficacy over the respective replication-deficient virus with GCV. However, GCV did not further enhance efficacy of Ad5/3-Δ24-TK-GFP in vivo. Simultaneous detection of tumor load and virus replication with bioluminescence and fluorescence imaging provided insight into the in vivo kinetics of oncolysis. In summary, TK/GCV may not add antitumor activity in the context of highly potent oncolysis.

Introduction

Virotherapy, utilization of the oncolytic properties of viruses for killing of tumor cells, is an attractive strategy for treating cancers resistant to conventional modalities.^{1, 2} Adenovirus, one of the most widely studied viral agents, can be genetically modified to selectively replicate in and destroy tumor cells through exploitation of molecular differences between normal and cancer cells. The life cycle of adenoviruses results in oncolysis of infected cells and spreading of virus progeny to surrounding cells for local amplification of input dose. Despite excellent preclinical data and good safety in humans, several obstacles remain.

First, the potency of oncolytic adenoviruses might be limited due to poor transduction of target cells. As the efficacy of replicating agents is dependent on their capability for infecting target cells, low receptor expression levels can make tumors refractory to oncolysis.³ Numerous previous studies have demonstrated that expression of coxsackie-adenovirus receptor (CAR), the primary receptor for adenovirus serotype 5 (Ad5), is frequently dysregulated in many advanced cancers.⁴ Various methods have been explored to circumvent CAR deficiency for increased infectivity of tumor cells and reduced normal tissue tropism.⁵ One strategy involves substituting the knob domain of the Ad5 fiber with the serotype 3 (Ad3) knob.⁶ This allows binding and entry through the Ad3 receptor, which is expressed at high levels in ovarian and many other tumors.^{7, 8, 9, 10}

Second, although the oncolytic process theoretically continues as long as target cells persist, clinical trials with early-generation replicating viruses have indicated that complete eradication of solid tumor masses is rarely seen. Although not much studied, preclinical evidence suggests that in vivo penetration of tumors is difficult due to intratumoral complexities.¹¹ A useful approach for improving the potency of replicating agents is to arm viruses with therapeutic transgenes such as genes encoding prodrug-converting enzymes.¹² This strategy has been useful with poorly replicating early-generation agents,^{13, 14} but has not been studied in the context of highly effective viruses. High oncolytic potency might reduce the utility of prodrug conversion, because of the strong bystander effect caused by virus dissemination per se. Also, prodrug conversion might affect virus replication, which might work against oncolysis.

In this study, we constructed an infectivity-enhanced selectively replicating adenovirus, Ad5/3-Δ24-TK-GFP, which contains a fusion gene between herpes simplex virus thymidine kinase (TK) and green fluorescent protein (GFP) in the partly deleted E3 region. TK is one of the most studied prodrug-converting enzymes utilized in gene therapy and it can convert prodrug ganciclovir (GCV) into a cytotoxic metabolite. This system is associated with killing of uninfected neighboring cells (bystander effect).¹⁵ Tumor selectivity of Ad5/3-Δ24-TK-GFP was achieved by engineering a 24 bp deletion in the retinoblastoma (Rb) binding site of E1A.^{16, 17} Therefore, Ad5/3-Δ24-TK-GFP replicates selectively in cells deficient in the Rb/p16 pathway. Defects of this pathway are present in most human tumors, including ovarian cancer.^{18, 19}

Results

Viral replication and correlation with the level of GFP expression

TK-GFP was inserted into the 6.7K/gp19K-deleted E3A region of a serotype 5 based oncolytic adenovirus featuring the serotype 3 fiber knob, resulting in a novel virus Ad5/3-Δ24-TK-GFP (Figure 1a). This virus utilizes endogenous E3 control elements for replication-coupled transgene expression approximately 8 h postinfection.^{20, 21} Ad5/3-TK-GFP, a virus with a CMV promoter driven transgene in the deleted E1 region, was constructed as a replication-deficient control.

Figure 1

Kinetics of Ad5/3-Δ24-TK-GFP in vitro. (a) Schematic representation of adenoviruses utilized in this study. All viruses are based on Ad5 but contain fibers modified with the Ad3 knob. In replication-deficient Ad5/3-TK-GFP, E1 is substituted with the TK-GFP fusion gene driven by a CMV promoter. All oncolytic viruses feature a 24 bp deletion in the Rb binding site of E1A for Rb/p16 pathway selective replication. In Ad5/3-Δ24-TK-GFP, TK-GFP is inserted into the 6.7K/gp19K-deleted E3A region under the control of the endogenous E3 promoter. Ad5/3-Δ24-Δgp19K, an isogenic control virus, contains a 6.7K/gp19K-deleted E3A region without a transgene. Ad5/3-Δ24 has an intact E3 region. (b) Ad5/3-Δ24-TK-GFP replication in ovarian cancer cells. SKOV3.ip1, MDAH 2774, OV-4 and Hey cells were infected with Ad5/3-Δ24-TK-GFP or control viruses and followed for cytopathic effect (CPE). CPE data was translated into infectious particles by standard TCID₅₀ calculations. Error bars indicate s.d.

To study if insertion of the fusion gene had an effect on virus replication, Ad5/3-Δ24-TK-GFP was compared to Ad5/3-Δ24-Δgp19K and Ad5/3-Δ24 (Figure 1b). The former is an isogenic virus featuring the 6.7K/gp19K deletion without a transgene and the latter contains an intact E3 region without a transgene. Ad5/3-Δ24-TK-GFP replicated efficiently on all analyzed cancer cell lines. However, replication of Ad5/3-Δ24-TK-GFP was slightly slower and total production of functional virions remained 1–2 log lower compared to the other replicating viruses. Interestingly, Ad5/3-Δ24-Δgp19K with a partly deleted E3 region resulted in higher replication levels than Ad5/3-Δ24, whose genome is closer to a wild-type adenovirus.

To confirm expression of TK-GFP fusion product, Ad5/3-Δ24-TK-GFP and Ad5/3-TK-GFP infected ovarian cancer cells were imaged daily for GFP expression (Supplementary Figure 1). With Ad5/3-TK-GFP, transgene production was evident at 1000 viral particles/cell (VP/cell) throughout the whole experiment (days 1–10 after infection) on MDAH 2774 cells. Ad5/3-Δ24-TK-GFP infected cells expressed GFP to higher degree than cells infected with the replication-deficient virus and the proportion of GFP-positive cells increased over time. By day 10, expression of GFP was seen in cells infected with only 0.001 VP/cell. Results were similar with other ovarian cancer cell lines (data not shown). These data suggest that higher amounts of fusion protein are produced from replicating than from E1 deleted viruses, and that the coupling of transgene expression to replication allowed detection of replication by analysis of GFP. As infections were carried out based on VP titer and Ad5/3-Δ24-TK-GFP had a similar but slightly higher ratio of VP/infectious particles (that is, slightly less infectious particles present in infection) in comparison to Ad5/3-TK-GFP, it is unlikely that the difference in GFP expression resulted from differences in functional titer.

Quantitation of viral replication in the presence of GCV

Next, the kinetics of Ad5/3-Δ24-TK-GFP were studied in the presence of GCV. SKOV3.ip1 monolayers were infected with 10 VP/cell of Ad5/3-Δ24-TK-GFP and 10 μg/ml GCV was added according to different schedules (Figure 2). When GCV was given 1 h postinfection (to ensure that it is present throughout viral DNA replication), practically no infectious particles were formed, indicating that GCV has an inhibitory effect on viral replication when given immediately after virus. Administration of GCV 24 h postinfection resulted in a 10-fold reduction in the amount of viable virus at time points 48 and 72 h after infection. A plateau was reached for Ad5/3-Δ24-TK-GFP after 48 h, and therefore it was logical that GCV added at this time point did not decrease the number of infectious particles. Longer follow-up was not possible in this setting because of cell death at 96 h postinfection due to the effects of virus and GCV.

Figure 2

Effect of GCV on Ad5/3-Δ24-TK-GFP replication. SKOV3.ip1 cells were infected with 10 VP/cell and 10 μg/ml GCV was added 1–48 h later. The amount of virus produced was determined at different time points after infection by TCID₅₀ method. Cell viability was retained for the duration of the experiment. Error bars represent s.d.

Oncolytic potency in combination with GCV on ovarian cancer cells

To evaluate if the TK/GCV system can improve the efficacy of oncolytic virotherapy in vitro, ovarian cancer cells were infected with Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP and exposed to GCV 1, 24, 48, 72 or 96 h postinfection. MTS assay was used as an indicator of cell viability. On SKOV3.ip1 cells, Ad5/3-Δ24-TK-GFP infection with 1 or 10 VP/cell in combination with GCV resulted in significantly enhanced cell killing compared to virus alone with all analyzed treatment schedules (P=0.02 when GCV was given 1 h after infection at 1 VP/cell, P<0.001 for all other time points) (Supplementary Figure 2). There was not much difference between different schedules. When GCV was given 48 h after infection with 1 VP/cell of Ad5/3-Δ24-TK-GFP, virus alone resulted in 87% cell viability whereas combination treatment decreased the amount of living cells to 10% (Figure 3a).

Figure 3

Oncolytic potency of Ad5/3-Δ24-TK-GFP on ovarian cancer cells with and without GCV treatment. (a) SKOV3.ip1, (b) Hey and (c) OV-4 cells were infected with indicated doses of Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP and exposed to 10 μg/ml GCV 48 h postinfection. Cell viability was measured with MTS assay and viability of uninfected cells was set as 100%. Error bars indicate s.d.

With Hey and OV-4 cells, replication of Ad5/3-Δ24-TK-GFP was rapid and administration of GCV 48 h postinfection did not significantly increase cytotoxicity (P=0.142 and P=0.186 at 1 VP/cell, respectively; Figures 3b-c). Other time points yielded similar data (not shown). GCV was useful in combination with replication-deficient Ad5/3-TK-GFP with all cell lines and schedules compared to virus alone (P<0.001 at 100 VP/cell). However, oncolytic virus alone led to more efficient cell killing than E1-deleted virus combined with GCV.

In summary, these data suggest that the utility of TK/GCV-mediated cell killing is most evident in the context of replication-deficient virus. In contrast to the effect on virus replication, GCV timing did not influence in vitro cell killing dramatically. However, it seems useful to allow the virus to replicate before GCV treatment because of the inhibitory effect on replication and subsequently reduced transgene production. SKOV3.ip1 cells are known to be quite sensitive to the bystander effect,²² which may explain why GCV increased cell killing despite abrogation of Ad5/3-Δ24-TK-GFP replication.

Effect on tumor growth

To study whether tumor growth in nude mice might be inhibited by utilizing Ad5/3-Δ24-TK-GFP coupled with TK/GCV system, SKOV3-luc tumor bearing mice were treated intratumorally with two cycles of Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP and GCV was administered intraperitoneally for one week starting 48 h after the last viral injections (Figure 4a). GCV alone did not influence tumor size (P=0.429). Both Ad5/3-Δ24-TK-GFP and Ad5/3-TK-GFP alone or together with GCV reduced tumor growth significantly compared to untreated animals (all P<0.006). Further, both Ad5/3-Δ24-TK-GFP alone or coupled with GCV were more efficient in inhibiting tumor growth than E1-deleted Ad5/3-TK-GFP with GCV (P<0.001). Interestingly, GCV did not provide additional therapeutic benefit when combined with Ad5/3-Δ24-TK-GFP (P=0.495) or Ad5/3-TK-GFP (P=0.891).

Figure 4

Efficacy of Ad5/3-Δ24-TK-GFP and TK/GCV on tumor growth and noninvasive bioluminescence imaging of cancer cells in a subcutaneous model of ovarian cancer. (a) Tumor size in nude mice treated with two rounds of 1 × 10⁸ VP Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP intratumorally on days 7–9 and 19–21. A dose of 50 mg/kg GCV was injected intraperitoneally 48 h after viral injections (days 11–17 and 23–29). Error bars indicate s.e.m. (b) Photons emitted by luciferase-expressing cancer cells at different time points after treatment. Error bars indicate s.e.m. (c) Pseudocolor images of subcutaneous tumors in flanks of mice. Color scale minimum and maximum have been adjusted so that they are identical in each image (For color figures see online version).

Bioluminescence imaging was utilized for detecting luciferase-expressing cancer cells and the total flux of emitted light (photons/s) was measured (Figure 4b). Pseudocolor images were created to visually demonstrate the growth of subcutaneous tumors (Figure 4c). Good correlation was seen between tumor size and photon emission data. Administration of GCV did not significantly affect photon flux when compared to viruses alone (P=0.899 for Ad5/3-Δ24-TK-GFP and P=0.061 for Ad5/3-TK-GFP).

Replication and spreading of TK-GFP-expressing viruses in flank tumors were followed by in vivo imaging of GFP expression (Figure 5). Signals emitted from Ad5/3-Δ24-TK-GFP treated tumors seemed higher compared to tumors injected with Ad5/3-TK-GFP, indicating effective replication of the virus, although differences were not statistically significant.

Figure 5

Replication of TK-GFP-expressing viruses in subcutaneous tumors. (a) Fluorescence emitted from Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP injected tumors on day 22 (24 h after last virus injection). Error bars indicate s.e.m. (b) Pseudocolor images of virus-induced GFP production in tumor cells. Color scale minimum and maximum have been adjusted so that they are identical in each image (For color figures see online version).

Antitumor efficacy in an orthotopic murine model of peritoneally metastatic ovarian cancer

Therapeutic efficacy of Ad5/3-Δ24-TK-GFP and feasibility of TK/GCV system was also evaluated in a murine model of ovarian cancer. Severe combined immune deficiency (SCID) mice with peritoneal SKOV3-luc carcinomatosis were injected intraperitoneally with Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP and GCV was given daily for 2 weeks starting 48 h after virus injection (Figure 6a). The median survival of untreated mice was 32 days and treatment with replication-deficient Ad5/3-TK-GFP together with GCV increased it to 36 days (P=0.001). Oncolytic Ad5/3-Δ24-TK-GFP alone or in combination with GCV resulted in median survival times of 46 and 47 days, respectively, which were both significantly improved when compared to untreated animals (P<0.001). However, GCV did not add a significant survival benefit in comparison to Ad5/3-Δ24-TK-GFP alone (P=0.481). Treatment with Ad5/3-Δ24-TK-GFP alone led to significantly enhanced survival compared to E1-deleted Ad5/3-TK-GFP combined with GCV (P=0.023).

Figure 6

The antitumor efficacy of Ad5/3-Δ24-TK-GFP and noninvasive bioluminescence imaging in a murine model of peritoneally disseminated ovarian cancer. (a) Survival of SCID mice treated with 1 × 10⁸ VP Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP intraperitoneally on day 5 followed by 50 mg/kg GCV intraperitoneally daily for 2 weeks starting on day 7. (b) Photons emitted by luciferase-expressing cancer cells at different time points after treatment. Error bars indicate s.e.m. (c) Pseudocolor images of intraperitoneal tumor burden. Color scale minimum and maximum have been adjusted so that they are identical in each image (For color figures see online version).

Intraperitoneal tumor growth was followed with bioluminescence imaging (Figure 6b). The extent of disease was visualized with pseudocolor images of luciferase expression (Figure 6c). Results were in accordance with survival data as photon flux from mice treated with Ad5/3-Δ24-TK-GFP alone or in combination with GCV did not differ significantly (P=0.707). However, both treatments resulted in a smaller tumor load than in untreated mice (P<0.002).

Discussion

The TK/GCV suicide system is a classic strategy utilized in gene therapy. The approach is based on delivery of a gene encoding a prodrug-converting enzyme. TK phosphorylates prodrug GCV to its corresponding monophosphate and cellular kinases further phosphorylate it into a toxic triphosphate form, which can incorporate into DNA. This can cause chain termination, DNA damage and ultimately cell death.¹⁵

Reports combining TK/GCV system with replicating adenoviruses have been rather controversial. There are papers demonstrating that antitumor efficacy of TK expressing oncolytic adenoviruses is enhanced by treatment with GCV.^{13, 14, 23} On the other hand, some studies suggest that GCV does not further improve the oncolytic potential of replicating adenoviruses, at least not in vivo.^{24, 25, 26, 27, 28} Interestingly, most studies resulting in enhanced efficacy with the TK/GCV system have been performed with viruses lacking E1B 55 kDa. It is well established that these viruses do not replicate as well as wild-type virus and consequently oncolytic potency is also low.^{29, 30, 31, 32, 33} In contrast, the Δ24 deletion (also known as dl922–947) does not negatively influence replication.^{16, 17}

To evaluate the utility of TK/GCV in the context of highly effective oncolysis, we developed Ad5/3-Δ24-TK-GFP, an oncolytic agent based on the Ad5/3-Δ24 platform, which is superior to wild-type virus with regard to killing of ovarian and other cancer cells in vitro and in vivo.^{34, 35, 36} The TK-GFP fusion gene was placed into the partly deleted E3 region under the control of endogenous adenoviral gene expression machinery. This insertion site has been previously demonstrated to be ideal for expression of transgenes because of tight linkage to virus replication.^{20, 21} Transgene production begins approximately 8 h postinfection, and a heterologous promoter or poly-A are not required. Both aspects are useful for retaining highest possible virus replication and space is saved for maximal transgene size.

Infection with Ad5/3-Δ24-TK-GFP resulted in complete killing of all analyzed ovarian cancer lines. However, Ad5/3-Δ24-TK-GFP replicated at slightly lower levels compared to Ad5/3-Δ24-Δgp19K and Ad5/3-Δ24, possibly due to transgene production competing for the cellular resources needed for amplification of the viral genome. Alternatively, because TK-GFP is 1.9 kB and therefore 1 kB longer than the deleted E3 fragment, larger genome size may have slowed replication somewhat. Slower replication was not caused by deletion of 6.7K/gp19K per se. In contrast, infection with the isogenic virus harboring the deletion but lacking a transgene resulted in faster replication and more efficient oncolysis compared to virus with intact E3, which underlines the impact of genome size on speed of adenovirus replication. Replication of Ad5/3-Δ24-TK-GFP was associated with increased GFP production in comparison to Ad5/3-TK-GFP.

It has been proposed that GCV might block viral proliferation and subsequent spreading of virus.^{23, 24, 25, 26} In this study, we saw dramatic inhibition of virus replication and oncolysis when GCV was administered immediately after Ad5/3-Δ24-TK-GFP infection. Such a safety switch mechanism might be useful if it becomes necessary to abrogate virus replication. However, inhibitory effects were not evident when GCV was given after a plateau for Ad5/3-Δ24-TK-GFP production was already reached, indicating the importance of GCV scheduling in order to achieve a balance between GCV-mediated decrease in cytotoxicity by inhibition of viral replication and GCV-mediated increase in cytotoxicity by bystander effect.

A significant increase in cytotoxicity was seen on SKOV3.ip1 cells when Ad5/3-Δ24-TK-GFP infection was followed by prodrug treatment, independent of timing of GCV administration. Therefore, despite abrogation of production of functional virions by GCV, TK-GFP was still produced and effectively mediated cell killing in these cells known to be quite sensitive to the bystander effect.²² The E3 promoter is an early promoter and therefore not compromised by lack of functional virion production.

GCV did not increase the cytotoxicity of Ad5/3-Δ24-TK-GFP on Hey or OV-4 cells. In a recent study, infection with another oncolytic adenovirus, Ad5-Δ24TK-GFP, resulted in enhanced cell killing of both Hey and OV-4 cells when combined with GCV.²⁸ Entry of Ad5-Δ24TK-GFP into target cells is dependent on CAR expression, which has been demonstrated to be low in Hey and OV-4 cells.⁷ Therefore, less cells were infected by the CAR binding Ad5-Δ24TK-GFP and therefore the need for bystander effect was more urgent. Further, expression of TK-GFP was driven by CMV promoter placed in the deleted E3 region, which might not be optimal for effective replication because both deletion of the entire E3 and CMV directed expression are known to reduce production of new virions. These findings suggest that the TK/GCV system has its greatest utility in constructs with low oncolytic potency.

The antitumor efficacy of Ad5/3-Δ24-TK-GFP followed by GCV treatment was evaluated in subcutaneous and intraperitoneal xenografts of ovarian cancer. Despite robust production of TK-GFP, as shown by in vivo GFP expression, GCV did not result in reduced tumor growth or increased survival when compared to oncolytic virus alone. This data is in accordance with previous studies.^{24, 25, 26, 27, 28} We utilized noninvasive bioluminescence and fluorescence imaging to follow tumor development and virus dissemination in living mice. Luciferase imaging correlated with tumor growth and survival, while GFP could be detected to demonstrate effective virus replication and spreading in tumor tissue.

In summary, Ad5/3-Δ24-TK-GFP is a useful agent for treatment of ovarian cancer because virus replication can be imaged by direct detection of GFP, or by indirect detection of TK with, for example, FIAU.³⁷ Our in vitro data indicates that virus replication can be inhibited by GCV without loss of the bystander effect for killing of uninfected cancer cells. Further studies are needed to resolve if there is a scenario where these two opposite effects of GCV would convert into a therapeutic or safety advantage. However, our in vivo data suggest that when viral oncolysis is effective, TK/GCV may not add to efficacy. Our conclusion is that other arming approaches may be more useful in the context of highly potent oncolytic adenoviruses.^{38, 39, 40, 41, 42}

Materials and methods

Cells and GCV

Human-transformed embryonal kidney cell line 293 was obtained from Microbix (Toronto, Canada), while human lung adenocarcinoma cell line A549 and human ovarian cancer cell line MDAH 2774 were purchased from ATCC (Manassas, VA, USA). Human ovarian adenocarcinoma cell lines SKOV3.ip1 and Hey were provided by Dr Janet Price and Dr Judy Wolf (both from MD Anderson Cancer Center, Houston TX, USA), and OV-4 was a kind gift from Dr Timothy J Eberlein (Harvard Medical School, Boston, MA, USA). Firefly luciferase-expressing ovarian adenocarcinoma cell line SKOV3-luc was provided by Dr Negrin (Stanford Medical School, Stanford, CA, USA). All cell lines were cultured in recommended conditions. GCV was purchased from Roche (Espoo, Finland). 50 mg/ml stock solution was prepared in sterilized water and the agent was further diluted in growth media immediately before use.

Construction of adenoviruses

To create oncolytic Ad5/3-Δ24-TK-GFP, we utilized a plasmid pTHSN-TGL. Briefly, plasmid was constructed by digesting pTHSN,²¹ a plasmid containing the E3 region of the adenoviral genome, with SunI/MunI, and inserting TK-GFP⁴³ into the resulting 965 bp 6.7K/gp19K deletion of E3A.²⁰ pAdEasy-1.5/3-Δ24-TGL was generated by homologous recombination in Escherichia coli BJ5183 cells (Qbiogene Inc., Irvine, CA, USA) between FspI-linearized pTHSN-TGL and SrfI-linearized pAdEasy-1.5/3-Δ24, a rescue plasmid containing the serotype 3 knob and a 24 bp deletion in E1A.²¹ The genome of Ad5/3-Δ24-TK-GFP was released by PacI digestion and subsequent transfection of A549 cells.

To construct the E1-deleted control virus Ad5/3-TK-GFP, we used PCR amplification to engineer BglII/XhoI restriction sites around the TK-GFP gene: 5′-ACAGATCTCTAGAGGATCTTGGTGGCGTGAA-3′ and 5′-TACTCGAGCTAGAGGATCCCCGGCCG-3′. pShuttle-CMV⁴⁴ was digested with BglII/XhoI and the TK-GFP fragment was ligated into the multiple cloning site under the control of the CMV immediate early promoter to generate pShuttle-CMV-TGL. Homologous recombination was performed between pAdEasy-1.5/3 and PmeI-linearized pShuttle-CMV-TGL to construct pAdEasy-1.5/3-TGL. The genome of Ad5/3-TK-GFP was released by PacI and transfected into 293 cells.

Ad5/3-Δ24-Δgp19K, an oncolytic virus containing a 965 bp deletion in E3A, was constructed as above, but without ligating TK-GFP into pTHSN. Ad5/3-Δ24, a virus with intact E3, has been previously described.³⁴

Viruses were amplified on 293 or A549 cells and purified on double cesium chloride gradients. The VP concentration was measured at 260 nm and infectious particles were determined by TCID₅₀ on 293 cells. The ratio of VP/infectious particles was 23, 10, 19 and 6 for Ad5/3-Δ24-TK-GFP, Ad5/3-TK-GFP, Ad5/3-Δ24-Δgp19K and Ad5/3-Δ24, respectively. Modified viral regions were confirmed by PCR and sequencing.

Viral replication and GFP expression on ovarian cancer cells

Cells were seeded at 1 × 10⁴ cells/well on 96-well plates and cultured overnight. Cells in 10 replicates were infected with viruses at 0.0001–1000 VP/cell in 100 μl of growth media with 2% fetal bovine serum (FBS). Cells were followed daily for cytopathic effect (CPE) for 10 days, and the amount of infectious particles was calculated based on CPE data by a modification of standard TCID₅₀ method. Cells were imaged daily with Typhoon 9400 variable mode imager (GE Healthcare Life Sciences, Helsinki, Finland) for expression of GFP, and images were analyzed with ImageQuant 5.20 (GE Healthcare Life Sciences).

Effect of GCV on viral replication

SKOV3.ip1 cells were plated at 5 × 10⁵ cells/well on six-well plates. Next day, cells were infected with Ad5/3-Δ24-TK-GFP at 10 VP/cell in 1 ml of growth media with 2% FBS for 1 h. Growth media was replaced with 3 ml of fresh 2% media or media containing 10 μg/ml GCV 1, 24 or 48 h after infection. Cells and media were harvested 24–72 h postinfection, lysed by three freeze-thaw cycles and centrifuged at 4000 r.p.m. for 10 min. The number of infectious particles present in the resulting supernatants was determined by TCID₅₀ on 293 cells.

Killing of ovarian cancer cells in vitro

Cells were seeded at 1 × 10⁴ cells/well on 96-well plates and cultured overnight. Cells in quadruplicate were infected with Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP at 0.01–100 VP/cell diluted in 100 μl of growth media containing 2% FBS. Growth media was substituted with 100 μl of 2% media with or without 10 μg/ml GCV 1 h, 24, 48, 72 or 96 h postinfection. Media was refreshed every third day and GCV was kept at 10 μg/ml. Cell viability was measured by using the CellTiter 96 AQ_ueous One Solution Cell Proliferation Assay (Promega, Madison, WI, USA) 7 days (SKOV3.ip1) or 9 days (Hey, OV-4) postinfection.

Murine models of ovarian cancer

Three- to four-week-old female Naval Medical Research Institute (NMRI) nude and C.B-17 SCID mice were obtained from Taconic (Ejby, Denmark) and quarantined for 2 weeks. Animal experiments were approved by the Experimental Animal Committee of the University of Helsinki and the Provincial Government of Southern Finland. Viruses and GCV were diluted in growth media without FBS, and untreated control mice received growth media only.

Subcutaneous xenografts were established by inoculating 5 × 10⁶ SKOV3-luc cells in both flanks of nude mice. When tumors reached a diameter of approximately 4–5 mm, mice were randomized into six treatment groups (n=8 tumors/group). A total of 1 × 10⁸ VP Ad5/4-Δ24-TK-GFP or Ad5/3-TK-GFP was delivered intratumorally in 50 μl of growth media on days 7, 8 and 9 after cell injection. Second round of viruses was given on days 19, 20 and 21. A dose of 50 mg/kg GCV was administered intraperitoneally daily starting 48 h after last virus injection during 1-week periods on days 11–17 and 23–29. Tumors were measured with a caliper and mice were killed when tumors reached a diameter of 10 mm. Tumor size was calculated by using a formula 0.5 × length × width². Mice were imaged noninvasively for luciferase and GFP expression on days 7, 15, 22 and 29 when possible.

An orthotopic model of ovarian cancer was developed by injecting 5 × 10⁶ SKOV3-luc cells intraperitoneally into SCID mice. On day 5, mice (n=9) were treated with 1 × 10⁸ VP Ad5/3-Δ24-TK-GFP or Ad5/3-TK-GFP intraperitoneally diluted in 500 μl of growth media. A dose of 50 mg/kg GCV was given intraperitoneally daily for 14 days starting 48 h after virus delivery. Mice were imaged noninvasively for luciferase expression on days 4, 8, 15, 22, 29, 36 and 43 when possible.

Noninvasive imaging

Mice were imaged using IVIS 100 (Xenogen, Alameda, CA, USA) to detect expression of luciferase or GFP. For bioluminescence imaging, 150 mg/kg D-luciferin (Promega, Madison, WI, USA) was injected intraperitoneally and images were captured 10 min later with 1.0 s exposure time, 1 f/stop, binning 2 and open filter. Fluorescence images were taken by using 1.0 s exposure, 2 f/stop, binning 2 and GFP filter. Photographic images were captured with 0.2 s exposure, 8 f/stop, binning 2 and open filter. Images were overlaid with Living Image 2.50 (Xenogen). Total flux (photons/s) was measured by drawing regions of interest (ROIs) around tumor areas enclosing emitted signals. Background was subtracted by measuring same sized ROIs in areas without light emission.

Statistical analysis

One-way analysis of variance with Bonferroni's post hoc test for multiple comparisons was used for statistical analysis of cell viability and tumor GFP expression data with SPSS 14.0 (SPSS Inc., Chicago, IL, USA). Analysis of tumor volume and photon emission was performed using a repeated measures model with PROC MIXED using SAS 9.1 (SAS Institute Inc., Cary, NC, USA). Measurements were log transformed for normality. The effects of treatment group, time in days and the interaction of treatment group and time were evaluated by F-tests. Curvature in the models was tested for by a quadratic term for time. The a priori planned comparisons of specific differences in predicted treatment means averaged over time and at the last time point were computed by t-statistics. Survival data was plotted into a Kaplan–Meier curve and groups were compared pair-wise with log-rank test using SPSS 14.0. For all analyses, P<0.05 was deemed statistically significant.