RESEARCH ARTICLE

Enhanced efficacy of Escherichia coli nitroreductase/CB1954 prodrug activation gene therapy using anE1B-55K-deleted oncolytic adenovirus vector

M-J Chen1,4, NK Green1,5, GM Reynolds2, JR Flavell3, V Mautner1, DJ Kerr1,6, LS Young1 and PF Searle1

1Cancer Research UK Institute for Cancer Studies, University of Birmingham, Edgbaston, Birmingham, UK; 2Liver ResearchLaboratories, Queen Elizabeth Hospital, Birmingham, UK; and 3Department of Pathology, University of Birmingham, Edgbaston,Birmingham, UK

Viruses that replicate selectively in cancer cells constitute anexciting new class of anticancer agent. The conditionallyreplicating adenovirus (CRAd) dl1520, which lacks the E1B-55K gene, has elicited significant clinical responses inhumans when used in combination with chemotherapy. Aconvergent development has been to use replication-defective viruses to express prodrug-activating enzymes incancer cells. This can sensitize the cancer to prodrug, butdepends upon achieving sufficient level, distribution andspecificity of enzyme expression within the tumour. In thisstudy, we have expressed the prodrug-activating enzymenitroreductase (NTR) in the context of an E1B-55K-deletedadenovirus, CRAd-NTR(PS1217H6). We show that CRAd-NTR(PS1217H6) retains oncolytic growth properties, and

expresses substantially more NTR than a comparable,replication-defective adenovirus. The combination of viraloncolysis and NTR expression results in significantly greatersensitization of SW480 and WiDr colorectal cancer cells tothe prodrug CB1954 in vitro. In vivo, CRAd-NTR(PS1217H6)was shown to replicate in subcutaneous SW480 tumourxenografts in immunodeficient mice, resulting in more NTRexpression and greater sensitization to CB1954 than withreplication-defective virus. Combination therapy of CRAd-NTR(PS1217H6) with CB1954 reduced tumour growth from13.5- to 2.8-fold over 5 weeks, and extended median survivalfrom 42 to 81 days, compared with no treatment.Gene Therapy (2004) 11, 1126–1136. doi:10.1038/sj.gt.3302271; Published online 27 May 2004

Keywords: adenovirus; CB1954; enzyme–prodrug therapy; cancer gene therapy; oncolytic virus; nitroreductase

Introduction

Viruses that undergo a lytic replication cycle selectivelyin malignant cells form a promising new class ofanticancer agent.1 The adenovirus (Ad) mutant dl15202

was engineered not to express the E1B-55K protein, animportant function of which is to inactivate the tumoursuppressor protein p53, thereby preventing p53-mediated cell cycle arrest or apoptosis during infectionof normal human cells and thus allowing efficient virusreplication. Ad dl1520, also known as Onyx-015, istherefore severely restricted in its ability to replicate innormal cells, whereas its replication in malignant cells, inwhich the p14ARF–p53 pathway is usually dysfunctional,is relatively unimpeded.3,4 Onyx-015 has been tested in anumber of clinical trials, involving single or multipledirect intratumoral injections, intravenous, intrahepaticarterial or intraperitoneal delivery to cancers of the headand neck, pancreas, ovary and liver metastases of

colorectal cancer; many of these trials were reviewedby Kirn.5 While the virus was shown to be safe by allthese routes of delivery, and at doses of up to 2� 1012

particles, efficacy as a single agent was low. However,when Onyx-015 was used in combination with che-motherapy, the modalities appeared to work synergisti-cally. A number of tumour nodules that had previouslyprogressed on chemotherapy alone demonstratedcomplete regression and extended time to recurrencefollowing virus injection and concurrent systemicadministration of cisplatin and 5-fluorouracil.5,6 Thecombination of chemotherapy with dl1520 virotherapydelivered via the hepatic artery also produced someresponses of colorectal cancer liver metastases that hadpreviously progressed on either chemotherapy or vir-otherapy alone.7

The localized activation of nontoxic prodrugs tocytotoxic derivatives by enzymes within a tumour hasthe potential to achieve superior antitumour efficacywithout the systemic, dose-limiting toxicity associatedwith conventional chemotherapeutic agents.8 A favouredmodality of cancer gene therapy involves using viralvectors to express prodrug-activating enzymes in tu-mour cells. Thus, expression of thymidine kinase fromherpes simplex virus (HSV-TK) sensitizes the infectedcells to the prodrugs acyclovir and ganciclovir (GCV);9

expression of cytosine deaminase from Escherichia coli oryeast sensitizes cells to 5-fluorocytosine (5-FC);10,11 and

Received 24 October 2003; accepted 21 February 2004; publishedonline 27 May 2004

Correspondence: Dr PF Searle, Cancer Research UK Institute for CancerStudies, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK4Current address: Mackay Memorial Hospital, Taipei, Taiwan5Current address: Hybrid Systems Ltd, Oxford BioBusiness Centre,Littlemore Park, Oxford OX4 4SS, UK6Current address: Department of Clinical Pharmacology, University ofOxford, Radcliffe Infirmary, Oxford OX2 6HE, UK

Gene Therapy (2004) 11, 1126–1136& 2004 Nature Publishing Group All rights reserved 0969-7128/04 $30.00

www.nature.com/gt

cells expressing E. coli nitroreductase (NTR) are sensi-tized to the prodrug CB1954 (5-[aziridin-1-yl]-2,4-dini-trobenzamide).12–14 Many of the reported studies withthese systems have employed replication-defective ade-novirus vectors, due to their high titre and efficient genedelivery to many cell types irrespective of cell prolifera-tion status. Each system exhibits a bystander effect,whereby localized spread of activated prodrug from cellsin which it is generated can kill other cells in the vicinitythat do not themselves express the enzyme. Never-theless, the limitations on the amount and distribution ofvirus delivery to intact tumours in vivo reduces theefficiency of tumour therapy in animal models belowthat which can be achieved when all tumour cellsendogenously express the enzyme.

There could be considerable benefits from combiningthese approaches, using a conditionally replicativeadenovirus to express prodrug-activating enzymes intumour cells. Relative to a replication-defective vector,the increased gene copy number due to virus replicationshould increase the expression of the prodrug-activatingenzyme selectively in the malignant cells, while thepotential for multiple cycles of spread beyond theinitially infected cells could improve the distribution ofvirus and prodrug-activating enzyme within the tumour.The demonstrated synergy of combining conventionalchemotherapy with oncolytic virotherapy predicts agreater benefit from combination of localized prodrugactivation with oncolytic virotherapy. Wildner et al15,16

have generated an E1B-55K-deleted adenovirus expres-sing HSV-TK. Intratumoral injection of this virus, incombination with GCV treatment, slowed the growth oftumour xenografts and prolonged survival of the mice.Freytag et al17 incorporated a CD/TK fusion gene in anE1B-55K-deleted adenovirus. In several tumour cell linesin vitro, addition of a single prodrug enhanced theoncolytic activity, although exceptions were noted. Invivo, addition of both prodrugs following virus admin-istration resulted in marked tumour regression andgrowth delay, which could be further enhanced bycombination with radiotherapy.18

The ultimate cytotoxic products generated from theprodrugs GCV or 5-FC are nucleotide analogues thatinhibit DNA synthesis. In contrast, activation of CB1954generates a potent DNA-crosslinking agent whosecytotoxicity is independent of ongoing DNA replica-tion.12,19,20 Treatment of mice with CB1954 can lead tocomplete regression of tumour xenografts that stablyexpress NTR following in vitro retroviral transduction21

or transfection.22 Tumour regression and prolongedsurvival have also been observed following in vivodelivery of replication-defective adenovirus expressingNTR and treatment with CB1954.20,23 However, theantitumour efficacy was suboptimal and dependent onvirus dose, which was limited by systemic toxicity. Weanticipated that use of a replicating vector could improvethe efficacy of the system, and report here the construc-tion and evaluation, both in vitro and in vivo, of an E1B-55K-deleted, conditionally replicating adenovirus ex-pressing NTR (CRAd-NTR(PS1217H6)).

Results

In CRAd-NTR(PS1217H6) (Figure 1), the NTR gene isinserted at the site of E1B-55K deletion, under the

transcriptional control of the CMV immediate-earlypromoter, and followed by an intron and 30-end proces-sing signals from human b-globin and complement C2genes, respectively, as previously described.20,23 ThisNTR expression cassette is similar to that in CTL102,23 afully E1-deleted, replication-defective adenovirus cur-rently in clinical trials.24 In CRAd-NTR(PS1217H6), weinserted the 30 end of the bovine growth hormone geneimmediately downstream of the E1B-19K coding regionto reduce possible transcriptional interference with theinserted transgene.

We also produced a similar construct, CRAd-EGFP(PS1217L), carrying the gene for enhanced greenfluorescent protein (EGFP) (Clontech) instead of NTR.This was used to check the infectability of the colorectalcarcinoma cell lines, SW480 and WiDr, infected atmultiplicities of 0.5, 5 or 10 plaque forming units(PFU)/cell. The percentage of cells expressing EGFPwas determined by flow cytometry 24, 48 and 72 h afterinfection (Table 1). The replication-defective virus, Ad-CMV-EGFP, was also used for comparison. This estab-lished that a multiplicity of 5–10 PFU/cell was sufficientto infect the majority of cells in these two lines.

Replication of CRAd-NTR(PS1217H6)The ability of CRAd-NTR(PS1217H6) to cause cytolysisof malignant and nontransformed cells as a result of viralreplication was compared with Ad5-WT. The two color-ectal carcinoma cell lines SW480 and WiDr, and the

Figure 1 Map of CRAd-NTR(PS1217H6). The E1A and E1B19K genesare derived from dl1520. bGH 30, 30 end of bovine growth hormone gene;CMV, cytomegalovirus immediate-early promoter; NR, NTR codingregion; bG/C2, splice sites and second intron from human b-globin and30-end signals from human complement C2 gene. The numbers correspondto the Ad5-WT sequence.

Table 1 Flow cytometric determination of EGFP-positive cells (%)following infection of colorectal carcinoma cell lines SW480 andWiDr with CRAd-EGFP(PS1217L) and Ad-CMV-EGFP

Virus MOI SW480 WiDr

24 h 48 h 72 h 24 h 48 h 72 h

CRAd-EGFP(PS1217L) 0.5 48 73 87 23 30 455 ND ND ND 84 92 96

10 91 96 98 ND ND ND

Ad-CMV-EGFP 0.5 59 68 79 38 43 615 ND ND ND 94 94 99

MOI: multiplicity of infection (PFU/cell); ND: not done.

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1127

Gene Therapy

nontransformed KS fibroblast and AC-3 keratinocytelines, all of human origin, were infected with the virusesat doses of 0, 0.1, 1, 10, 50 or 100 PFU/cell, and observeddaily. The cultures infected with 1 PFU/cell Ad5-WTshowed 480% destruction of the cell sheet of SW480,WiDr, KS and AC-3 cells at 8, 7, 19 or 15 days followinginfection, respectively, at which times the cultures werestained with crystal violet (Figure 2a). In the tumour celllines, oncolysis equivalent to Ad5-WT at 1 PFU/cell wasobserved with 10 PFU/cell of CRAd-NTR(PS1217H6).In the normal KS fibroblasts, equivalence was obtainedwith 50 PFU/cell CRAd-NTR(PS1217H6), while in theAC3 keratinocytes even 100 PFU/cell of CRAd-NTR(PS1217H6) caused only partial lysis of the cellsheet.

In a separate experiment, SW480, KS and AC-3 cellswere infected with 0, 0.1, 1, 10 or 100 PFU/cell of CRAd-NTR(PS1217H6). All were stained with crystal violetwhen the SW480 cultures infected with 1 PFU/cellshowed complete oncolysis, 14 day after infection. Atthis time, the KS cell cultures were eliminated at100 PFU/cell, but showed no growth inhibition at10 PFU/cell. The AC-3 keratinocytes showed just partialcytolysis at 100 PFU/cell (Figure 2b).

The time course of virus DNA replication wasmonitored following infection of SW480 cells with0.5 PFU/cell of CRAd-NTR(PS1217H6). DNA was ex-tracted, dot-blotted and probed with a 32P-labelled probefrom the E4 region of the virus. As shown in Figure 3a,virus DNA was below the threshold of detection at 12 h,but gave a distinct signal at 24 h with increasing intensityat 36 and 48 h. No signal was detected at any time pointfrom parallel cultures infected with the E1-deleted,replication-defective adenovirus CTL102. The signal forCRAd-NTR(PS1217H6) was quantitated using a phos-phoimager, in comparison with a standard curve. TheDNA levels at 24, 36, 48, 72, 96 and 120 h correspondedrespectively to 37715, 13037280, 28217393, 439371242,29437563 and 26567451 copies of the virus genome perinfected cell (mean7s.d. of duplicates).

2 12 24 36 48 72 96 120 h

100 10 1 0.1 ng

CRAd-NTR

CTL102

standard

CRAd-NTR

CRAd-NTR

CTL102

CTL102

WiDr

SW480

0 24 48 72 96 1200.001

0.01

0.1

1

10

100

1000

Time after infection (hours)

0 24 48 72 96

0.001

0.01

0.1

1

10

100

1000

Time after infection (hours)

a

b

c

Figure 3 Time course of virus replication. (a) Dot-blot of virus DNAreplication. SW480 cells (1�106) were infected with 0.5 PFU/cell ofCRAd-NTR(PS1217H6) or CTL102 as indicated. Total DNA wasextracted at the indicated times, dot-blotted and hybridized with a DNAprobe from the E4 region, present in both viruses. (b) Cultures of 1�105

SW480 and (c) WiDr cells were infected with 10 PFU/cell of CRAd-NTR(PS1217H6) or CTL102. At intervals, cultures were harvested andthe yield of infectious virus particles was determined by plaque titration on911 cells.

Figure 2 Dose, time and host cell dependence of cell lysis by CRAd-NTR(PS1217H6). (a) Cultures of SW480 and WiDr colorectal carcinoma,KS fibroblasts and AC3 keratinocytes were infected with CRAd-NTR(PS1217H6) or Ad5-WT at the indicated doses. Cultures were fixedand stained with crystal violet at the time indicated after infection, whenthe culture infected with 1 PFU/cell Ad5-WT showed essentially completelysis. (b) Cultures of SW480, KS and AC3 cells were infected with theindicated doses of CRAd-NTR(PS1217H6). Cultures were fixed andstained with crystal violet 14 days after infection, when the SW480 cellsinfected with 1 PFU/cell showed essentially complete lysis of the cellmonolayer.

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1128

Gene Therapy

To monitor the generation of new infectious virusparticles, plaque assays using 911 cells were performedon cell lysates at intervals following infection of SW480cells with 10 PFU/cell CRAd-NTR(PS1217H6) or CTL102(Figure 3b). The titre of CRAd-NTR(PS1217H6) increasedabout 104-fold from its nadir at 4 h, with most of theincrease (almost 103-fold) occurring between 24 and 48 hpost infection. The peak virus yield of 177713.5 PFU/cell corresponds to 18-fold amplification of the inputdose. The titre of infectious CTL102 in contrast showeda slight rise between 4 and 24 h, but the peak levelcorresponds to o1% of the input dose; by 96 h, thelevel was 4104-fold lower than that of CRAd-NTR(PS1217H6). Similar results were obtained in WiDrcells, in which CRAd-NTR(PS1217H6) reached a max-imum of 568722.6 PFU/cell (Figure 3c), or 450-foldamplification of the input virus.

Comparison of NTR expressionAn expected benefit of a replicating vector is that theincreased gene copy number could lead to a substantialincrease in the level of transgene expression, comparedto the equivalent replication-defective vector. To confirmthis and establish the time course, total protein sampleswere prepared from cultures of SW480 or WiDr cellsinfected with 10 PFU/cell of CRAd-NTR(PS1217H6) orCTL102, and levels of NTR were monitored by Westernblot. As shown in Figure 4, NTR was just detectable at12 h following infection with CRAd-NTR(PS1217H6),and increased rapidly to a strong maximum signal by48 h. Levels of NTR expressed by CTL102 increasedmuch more gradually throughout the time course; inSW480 cells, the signal at 24 h was comparable to thatfrom CRAd-NTR(PS1217H6) at 12 h post infection, andeven by 96 h the expression from CTL102 was muchweaker than that from CRAd-NTR(PS1217H6) at 24 h.Similar results were obtained in WiDr cells.

Effect of prodrug activation on virus replicationWe investigated the potential impact of CB1954 exposureon the generation of infectious progeny virus. SW480cells were infected with 10 PFU/cell CRAd-NTR(PS1217H6), then exposed to 2 or 10 mM CB1954

commencing 2, 24 or 48 h after infection. The cultureswere harvested at 72 h post infection, and the titre ofCRAd-NTR(PS1217H6) was determined by plaque assay.As shown in Figure 5, the yield of infectious virus wasreduced by about 50- or 500-fold by 2 or 10 mM CB1954,respectively, when prodrug was added 2 or 24 h afterinfection. In these conditions, the recovered virus wasless than the input dose. Delaying addition of prodruguntil 48 h post infection caused smaller reductions intitre relative to control cultures not treated with CB1954,and a modest increase over the input dose (1.5- and 4-fold, at 10 or 2 mM CB1954, respectively). This could bedue in part to the shorter overall duration of CB1954exposure, but also correlates with the time course ofreplication shown in Figure 3, in which the peak periodof virus replication is 24–48 h post infection. We concludethat to obtain optimum benefit of viral replication inaddition to the suicide gene therapy, addition of prodrugshould be delayed until at least 48 h following infection.

Cell killing by combination of prodrug activation withviral oncolysisSW480 cells were infected with CRAd-NTR(PS1217H6)at 5 PFU/cell, or CTL102 at 5 and 100 PFU/cell. After48 h, CB1954 was added to the cultures at a range ofconcentrations, and cell viability was determined 4 dayslater by MTS assay. As shown in Figure 6a, CTL102 at5 PFU/cell resulted in at most 10% reduction in viablecell number, at the highest prodrug concentration(100 mM CB1954); this was similar to the killing ofuninfected cells at this prodrug concentration (notshown). The oncolytic effect of CRAd-NTR(PS1217H6)in the absence of prodrug resulted in 11% reduction inviable cells. The combination with prodrug resulted in47, 71 and 93% loss of cell viability at 5, 10 and 20 mM

CB1954, respectively. These levels of prodrug-dependentcell killing were greater than achieved using a 20-foldhigher dose of the replication-defective virus CTL102.However, CTL102 was more effective in combinationwith CB1954 than the parental oncolytic virus dl1520; in aseparate experiment, 490% of SW480 cells infected withCTL102 at 20 or 50 PFU/cell were killed at 100 mM

CB1954 (50% cell killing at 25 and 5 mM CB1954,

CRAd-NTR

CTL102

CRAd-NTR

CTL102

NTR U 12 24 48 72 96 h

NTR U 12 24 48 72 120 h96 WiDr

SW480

Figure 4 Time course of NTR expression. SW480 and WiDr cells wereinfected with 10 PFU/cell CRAd-NTR(PS1217H6) or CTL102. Cells wereharvested at the indicated time points and 50 mg total cell protein wasanalysed by denaturing electrophoresis and Western blot to show NTR.NTR, bacterially expressed and purified NTR; U, uninfected cells.

Figure 5 Inhibition of CRAd-NTR(PS1217H6) replication by CB1954.SW480 cells (1�105 per well) were infected with 10 PFU/cell CRAd-NTR(PS1217H6) and then exposed to 2 or 10 mM CB1954 at 2, 24 and48 h after virus infection. All cultures were harvested 72 h after infectionand virus titres were determined by plaque assay on 911 cells (mean7s.d.of duplicate samples).

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1129

Gene Therapy

respectively) whereas the equivalent doses of dl1520killed only B20–45% of the cells at 100 mM CB1954(results not shown). Thus the enhanced cell killing by thecombination of CRAd-NTR(PS1217H6) and CB1954 isdue to the expression of NTR able to activate theprodrug.

In WiDr cells (Figure 6b), almost 40% reduction incell viability was seen with 5 PFU/cell CRAd-NTR(PS1217H6) in the absence of prodrug; this increasedto B70% reduction with 20 mM CB1954, whereas 5 PFU/cell CTL102 caused only 13% reduction in viability at20 mM CB1954. The combined effects of oncolysis andprodrug activation with CRAd-NTR(PS1217H6) at5 PFU/cell were greater than the cell killing caused byCB1954 plus CTL102 at 25 PFU/cell (Figure 6b).

In vivo studies with CRAd-NTR(PS1217H6)To investigate the replication of CRAd-NTR(PS1217H6)in vivo, subcutaneous SW480 tumour xenografts wereestablished in nude mice. When the tumours reached 5–8 mm in diameter, they were injected with 2� 108 PFU ofeither CRAd-NTR(PS1217H6) or CTL102. Tumours wereharvested 12 and 72 h later and analysed for replicationof vector DNA and for NTR expression. Virus DNA wasanalysed by quantitative PCR. As shown in Figure 7a,low and variable levels of virus DNA were detected at12 h; however, the mean level of CRAd-NTR(PS1217H6)was slightly higher than CTL102 (0.2870.28 versus0.170.09 ng/tumour, corresponding to B7.2 and2.8� 106 genome copies, respectively). By 72 h afterinjection, the level of CRAd-NTR(PS1217H6) DNA hadincreased B500-fold to 138713.8 ng (B3.8� 109 copies).

More surprisingly, even the E1-deleted CTL102 virusDNA had increased B85-fold, to 8.4276.46 ng DNA(B2.3� 108 copies). This is unlikely to be due tocontaminating, replication-competent adenovirus (RCA)since no cytopathic effects were observed, and minimalreplication could be measured, in vitro (see above). Also,quantitative PCR analysis of the stocks of CTL102established an upper limit of one copy of E1A sequence(potentially representing RCA) per 0.9–5.3� 105 virusparticles, which could not account for the level ofreplication detected.

It has been reported previously that at high multi-plicities of infection, E1-deleted adenoviruses can repli-cate their DNA in a variety of human tumour cell lines,in a dose- and cell cycle-dependent manner.25 This effectwould be negligible in our in vitro experiments, whichused a maximum of 10 PFU/cell; however, followinginjection of 2� 108 PFU into a tumour, inevitably cellsclose to the injection site may be infected at much highermultiplicities, facilitating such E1-independent virusreplication. Nevertheless, the CRAd replicated to levelsB15-fold higher than CTL102 in the tumours.

NTR expression in the tumours was monitored byWestern blotting (Figure 7b). A detectable level of NTRwas produced in the CTL102-injected tumour at 12 h,and the intensity of the signal increased by 72 h afterinjection. The signal from tumours injected with CRAd-NTR(PS1217H6) also increased substantially between 12and 72 h after injection, and even at 12 h the level of NTRwas greater than that expressed by CTL102 at 72 h.

12 1272 72 h

CRAd-NTR CTL102

NTR PBS

anti-NTR

anti-actin

a

b

12 72

0.01

0.1

1

10

100

CTL102

CRAd-NTR

Time (h)

Figure 7 Replication and NTR expression in SW480 tumours, 12 and72 h after infection with CRAd-NTR(PS1217H6) or CTL102. (a) VirusDNA per tumour was measured by Q-PCR. (b) Western blot of 100 mgprotein samples, probed using primary antibodies against NTR or actin (asa loading control).

Figure 6 Combination oncolytic and prodrug activation cytotoxicity. (a)SW480 cells and (b) WiDr cells (5000 per well of a 96-well plate) wereinfected with CRAd-NTR(PS1217H6) or CTL102 at the indicated doses.After 48 h, CB1954 was added at a range of concentrations, and cellsurvival was determined by MTS assay after a further 4 days. Points showmean7s.d. of triplicate wells.

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1130

Gene Therapy

NTR expression was also monitored by immunohis-tochemistry in SW480 tumours injected with 1�108 PFUof CRAd-NTR(PS1217H6) and CTL102 (Figure 8). At 3days after injection, both viruses gave comparable areasof the tumour sections in which many of the tumour cellsstained positively for NTR expression (Figure 8a and e;higher magnification in Figure 8b and f). While quanti-tative comparisons are difficult, the intensity of cell

staining appears greater for tumours injected withCRAd-NTR(PS1217H6) (Figure 8e and f) than withCTL102 (Figure 8a and b). By 7 days after virus injection,NTR expression in tumours injected with CTL102appears substantially reduced from day 3 with relativelyfew, scattered cells showing convincing NTR expression(Figure 8c and d). On day 7, there appears markedlymore NTR expression in tumours injected with CRAd-NTR(PS1217H6) than with CTL102, although this maystill represent a decline since day 3 (Figure 8g and h); it ispossible that viral oncolysis could have contributed tothis decline. Adenovirus structural proteins, which areonly expressed late in the replication cycle, were alsomonitored; Figure 8j shows strong staining for adeno-virus structural proteins in tumour cells 7 days afterinjection of CRAd-NTR(PS1217H6), which colocalizedwith expression of NTR in adjacent sections (Figure 8i).Adenovirus proteins were not detectable in tumoursinjected with CTL102 (not shown). Thus, although theinitial distribution of NTR expression within the tumourswas similar with both the replication-defective andconditionally replicative viruses, the combined data fromWestern blotting and immunohistochemistry indicatethat considerably more NTR is produced, and theduration of high expression is greater, in tumoursinfected with the oncolytic CRAd-NTR(PS1217H6) thanin those infected with CTL102.

To compare the therapeutic efficacy of CRAd-NTR(PS1217H6) with CTL102, subcutaneous SW480tumour xenografts 5–8 mm in diameter were injectedon each of 5 consecutive days with 2� 108 PFU of theviruses, or PBS as a control. This regimen was based onthe findings of Heise et al26 that injecting Onyx-015(dl1520) on 5 successive days was more effective than thesame total dose delivered on a single occasion. Subse-quently, each group of mice was further subdivided,with half of the animals receiving intraperitoneal injec-tions of CB1954 (25 mg/kg) on 4 successive days, theothers receiving the PBS vehicle. Prodrug treatment wascommenced 48 h after the last virus injection, to allowmaximum NTR expression even from the last virusinjection (see Figure 4). Although Figure 5 suggests thiswill reduce the potential yield of infectious progenyvirus, at least from the last injection, Figure 3 suggestedthe replication cycle should be almost complete, and theinterval since the earlier virus injections is correspond-ingly longer (up to 6 days).

Figure 9 shows the average tumour volumes for eachexperimental group plotted over the 35 days followingthe initial virus injections. The control, untreatedtumours grew at the fastest rate over this time, so thatby day 35 several of the mice had to be culled due totumour size; the average tumour size (7s.e.m.) of thisgroup at 35 days was 16487227 mm3. The group treatedwith CB1954 alone showed some reduction in tumourgrowth, (10177146 mm3 on day 35, P¼ 0.0375). Tumoursinjected with the replication-defective virus CTL102 hadan average volume of 7847210 mm3 in the absence ofprodrug, which was reduced to 6127103 mm3 for thegroup also treated with CB1954. Tumours injected withCRAd-NTR(PS1217H6) had average sizes of 6177165and 311754 mm3 in the absence and presence of CB1954treatment, respectively. The reductions in tumourvolume caused by both CTL102 and CRAd-NTR(PS1217H6) in the absence of prodrug were

Figure 8 Immunohistochemistry for NTR and adenovirus capsid proteins.Subcutaneous SW480 colon cancer xenografts were injected with1�108 PFU CRAd-NTR(PS1217H6) or CTL102. Tumours were har-vested after 3 or 7 days for immunohistochemistry, and photographed at� 100 (a, c, e, g) or � 200 (b, d, f, h–j) magnification. (a, b) Expression ofNTR 3 days after injection of CTL102. (c, d) Expression of NTR 7 daysafter injection of CTL102. (e, f) NTR expression 3 days after injection ofCRAd-NTR(PS1217H6). (g–i) NTR expression 7 days after injection ofCRAd-NTR(PS1217H6). Panels (b, d, f, h) show details of the field of viewshown in panels (a, c, e, g), respectively. (i, j) Serial sections showingcolocalization of NTR and adenovirus capsid proteins, respectively, 7 daysafter injection of CRAd-NTR(PS1217H6).

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1131

Gene Therapy

significant (P¼ 0.0149 and 0.0032, respectively), andthe combination treatments of CTL102 or CRAd-NTR(PS1217H6) with CB1954 both caused significantreductions in tumour growth, relative to CB1954 alone(P¼ 0.0425 and 0.0007, respectively). The oncolytic virusCRAd-NTR(PS1217H6) was significantly more effectivethan CTL102 in the presence of CB1954 (P¼ 0.0238).

The survival data are shown in Figure 9b. The mediansurvival times for groups injected with PBS, CTL102 orCRAd-NTR(PS1217H6) were 42, 75 and 78 days respec-tively without CB1954, and 62, 72 and 81 days respectivelywith CB1954 treatment. The prolonged survival dueto CB1954 without either virus was significant (P¼ 0.01).In the absence of prodrug, both viruses significantlyprolonged survival compared with untreated tumours(P¼ 0.0009 and 0.003 for CTL102 and CRAD-NTR(PS1217H6), respectively), while among the CB1954-treated groups, the increased survival mediated byCRAd-NTR(PS1217H6) was significant (P¼0.0086) whereasthat mediated by CTL102 was borderline (P¼ 0.0896).

Discussion

Clinical trials are underway of CTL102, a replication-defective adenovirus expressing NTR from the CMV

promoter, in combination with CB1954.23,24,27–29 Relativeto this virus, CRAd-NTR(PS1217H6) was expected todeliver improved efficacy due to increased expression ofNTR, improved spread within the tumour, and synergybetween viral oncolysis and the cytotoxicity of activatedprodrug.

Initial studies investigating the replicative propertiesof CRAd-NTR(PS1217H6) showed it to be stronglyattenuated relative to Ad5-WT in nontransformed hu-man cells, requiring around 50–100 times higher infec-tion doses to achieve equivalent lysis of primaryfibroblasts (KS) and AC-3 keratinocytes. In the colorectalcancer cell lines SW480 and WiDr, although CRAd-NTR(PS1217H6) is still attenuated relative to Ad5-WT,the differential is no more than 10-fold. Thus, as with theprototypic E1B-55K-deleted adenovirus dl1520, CRAd-NTR(PS1217H6) is oncolytic, that is, demonstrates theability to replicate and spread in cancer cells while beingstrongly attenuated in normal or nontransformed cells.Although the molecular basis for this effect has tran-spired to be more complex than was initially pro-posed,4,30,31 the established safety record of dl1520 inclinical trials commended this basic design of the vector.An additional safety feature is that insertion of the CMV-NTR expression cassette at the site of E1B-55K deletionensures that if, by recombination with E1 sequences incomplementing cell lines or Ad5-WT, the virus were toacquire a functional E1B-55K gene, and hence fullreplication-competence, this would lead to concomitantdeletion of the transgene, preventing the generation of afully replication-competent virus carrying the transgene.

By 72–96 h following infection of SW480 and WiDrcolorectal carcinoma cell lines in vitro with CRAd-NTR(PS1217H6), the titre of infectious virus rose by 3–4orders of magnitude, consistent with the observedincrease in virus DNA and corresponding to B18- to57-fold amplification of the input virus. Similarly in vivo,a large increase in the amount of CRAd-NTR(PS1217H6)DNA was observed between 12 and 72 h followingintratumoral injection of the virus. In vivo, the DNA ofE1-deleted vector CTL102 replicated to a level approxi-mately one-fifteenth that achieved by CRAd-NTR(PS1217H6). Although unanticipated, replication ofE1-deleted vectors has been reported previously in avariety of tumour cell lines; this depended on thenumber of viral genomes present in the cell nucleus,but standardized infection conditions used from 50 to1500 PFU/cell, for different cell lines.25 Although our invitro experiments used a maximum of 10 PFU/cell,intratumoral injection of 2� 108 PFU would result inmany cells taking up sufficient CTL102 to allow such E1-independent replication. It is interesting to speculate thatthis could be responsible for the prodrug-independentantitumour effects of CTL102 seen in Figure 9. Never-theless, the replication of CRAd-NTR(PS1217H6) wassignificantly greater, and able to occur at low multi-plicities of infection, accounting for the oncolytic actionof this virus in vitro and its greater efficacy at slowingtumour growth in vivo.

As predicted, CRAd-NTR(PS1217H6) gave substan-tially higher levels of NTR expression relative to CTL102both in vitro and in vivo. In vitro, 5 PFU/cell of CRAd-NTR(PS1217H6) resulted in greater sensitization ofSW480 cells to CB1954 than 100 PFU/cell of CTL102.The increased NTR expression from the replicating

Figure 9 In vivo therapeutic model. Subcutaneous SW480 tumours5–8 mm in diameter in nude mice were injected intratumorally with2� 108 PFU CTL102 or CRAd-NTR(PS1217H6), or PBS as control, for 5successive days. After further 2 days, the mice were injected with CB1954(25 mg/kg/day) for 4 successive days, or PBS as control (seven mice pergroup). (a) Mean tumour volume (7s.e.) over 35 days from first virusinjection. (b) Kaplan–Meier survival analysis.

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1132

Gene Therapy

vector and increased sensitization to CB1954 were alsoseen in WiDr cells, although in this case viral oncolysisalone caused significant cell loss within the timescale ofthe experiment. When the data are normalized relative tocell survival in the presence of the virus but absence ofprodrug, CRAd-NTR(PS1217H6) provided sensitizationto CB1954 comparable to a five-fold greater dose ofCTL102 in WiDr cells. Thus enhanced killing of bothtumour cell lines was obtained by the combination ofCB1954 prodrug activation and oncolytic replication ofthe vector. Preliminary data (not shown) suggest thatnontransformed cells are relatively resistant to thecombination of CRAd-NTR(PS1217H6) and CB1954.

In vivo, the combination of CRAd-NTR(PS1217H6) andCB1954 was the most effective treatment for subcuta-neous SW480 tumour xenografts, the treated tumourvolume growing just 2.8-fold over the 5 weeks followingstart of treatment, compared to an increase of 13.5-fold inuntreated tumours over the same period. Treatment withCB1954 alone caused a modest reduction (8.5-foldgrowth); this might be due to expression of DTdiaphorase, which is often upregulated in humancancers and can activate CB1954, albeit inefficiently.32

Tumours treated with CRAd-NTR(PS1217H6) withoutprodrug grew five-fold, while tumours treated withCTL102 plus prodrug grew 5.4-fold. This benefit ofcombining viral oncolysis with CB1954 activation wasalso reflected in the survival analysis of the mice, withanimals receiving CRAd-NTR(PS1217H6) and CB1954having the longest median survival, and being the onlytreatment group to achieve significantly longer survivalthan mice treated with CB1954 alone.

Both HSV-TK and a CD/TK fusion gene havepreviously been shown to improve the antitumoralefficacy of E1B-55K-deleted oncolytic adenoviruses,when used in conjunction with their respective prodrugs.In xenograft models using HT29 colon carcinoma, A375melanoma and ME180 cervical carcinoma, the oncolyticvirus expressing TK alone provided B1.5- to 3-foldincrease in median survival, while combination withGCV treatment (initiated 3 days after virus injection)increased the differential to B3.5- to 45-fold comparedto untreated controls.15,16 In C33A cervical carcinomaxenografts, the oncolytic adenovirus expressing the CD/TK fusion gene gave a 20-day growth delay, which wasnot significantly extended by either prodrug alone, butthe combination with both GCV and 5-FC extended thisB3-fold.18 Additional use of radiotherapy resulted in afurther marked improvement in growth delay.18 Thus theprinciple of improved efficacy from combining oncolyticviruses with prodrug activation, already established forantimetabolite prodrugs, has been extended to analkylating agent prodrug, and it may be anticipated thatfurther combinations between CRAd-NTR(PS1217H6)/CB1954, and TK/GCV, CD/5-FC, or radiotherapy couldshow further benefit. Our previous work suggests thatNTR/CB1954-induced cell death also contributes to anantitumour immune response and such an effect couldbe further enhanced in the context of an oncolytic virus,providing an additional immune bystander effect.33

CB1954 exposure of cells infected with CRAd-NTR(PS1217H6) led to a reduction in the yield of virusprogeny (Figure 5). Similar effects have also beenobserved on the replication of CRAds expressing TK orCD, when treated with their respective prodrugs. These

effects are not unexpected, and are indeed a safetyfeature, allowing termination of virus replication ifrequired.16,17 Clearly, the timing of prodrug administra-tion will be a key factor in the overall efficacy;administration too early in the course of oncolyticinfection could be counterproductive, preventing virusreplication and thus limiting both the oncolyic effect, andthe amount and distribution of the prodrug-activatingenzyme. However, with optimum timing, prodrugactivation is expected to kill more cells, or do so morerapidly while the tumour is smaller, than could beachieved by viral oncolysis alone. Indeed, Figure 9 showsa clear benefit in combining the modalities. Theoretically,prodrug activation might even enhance spread of theoncolytic virus, if direct and bystander cell killingallowed more efficient release and dissemination ofprogeny virus, even if their number were reduced.

In principle, the expression of prodrug-activatingenzymes from a late viral promoter, to delay theirexpression until after the onset of DNA synthesis, couldlimit the extent to which prodrug activation depressesthe virus yield.34,35 However, if infection of nearby cellswas asynchronous, viral replication might still beinhibited through a ‘bystander effect’, due to localspread of activated prodrug from cells that were laterin the infection cycle and thus contained sufficientactivating enzyme. Furthermore, from a safety view-point, such vectors have the disadvantage that thetransgene could be transmitted to a fully replication-competent virus by homologous recombination.

The E1B-55K-deleted adenoviruses are considerablyattenuated relative to Ad5-WT even in permissive cancercells, an effect that may be due to the role of E1B-55K inviral late RNA transport.36 Wildner and Morris37 foundthat expression of E1B-55K improved the oncolyticactivity of their HSV-TK-expressing vectors; however,combination with GCV now decreased the antitumourefficacy, apparently due to interference with viralreplication. In another study, expression of HSV-TK inan adenovirus with a WT E1 region did further delaygrowth of glioma xenograft tumours and prolongedsurvival in combination with GCV, compared with thevirus alone.35 Because these viruses expressed the fullcomplement of WT E1 genes in a nonrestricted manner,their replication is less selective for cancer cells than E1B-55K-deleted adenoviruses raising potential safety issues.However, natural adenovirus infections are generallycontrolled effectively by the immune system, and theexpressed HSV-TK gene would allow termination ofreplication in vivo with GCV, if viral replication outsidethe tumour approached dangerous levels. Heise et al38

demonstrated that an adenovirus with a deletion of thepRB binding pocket of E1A is at least as oncolytic asAd5-WT in cancer cells but markedly attenuated innormal cells, while adenoviruses in which expression ofWT E1 proteins is controlled by tumour-specific promo-ters could be as oncolytic as WT virus in cancer cells butas replication defective as E1-deleted viruses in allnormal cells, depending upon the selectivity providedby the promoter.39–41 Thus it is clear that replicatingvectors can deliver greater therapeutic efficacy thanreplication-defective vectors for virus-directed enzymeprodrug therapy. While in principle the ideal oncolyticvirus might be fully efficacious as a single agent,multiple enzyme–prodrug combinations have been

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1133

Gene Therapy

shown to enhance the overall efficacy in a number ofexperimental models, as anticipated from the therapeuticbenefit of combining multiple agents in conventionalcancer therapy. Owing to both direct cell kill and thebystander effect from prodrug activation, such combina-tion therapy may help to overcome limitations ononcolytic vectors in cancer patients due to heterogeneityin the infectability of tumour cells or eventual termina-tion of virus spread due to immune or other mechan-isms.

Materials and methods

CellsHuman colorectal carcinoma cell lines SW480 and WiDr,primary human skin fibroblast (KS)42 and 911 cells(human embryonic retinoblast transformed with adeno-virus E1 region)43 were cultured in DME/HEPES with10% of fetal calf serum and 2 mM glutamine. AC-3keratinocytes (a subclone of SCC12F that is immortalizedbut otherwise retains the growth characteristics ofnormal cells)44 were cultured in DME/HEPES containing5% fetal calf serum, 2 mM glutamine and 0.4 mg/mlhydrocortisone. All cells were kept at 371C in a humidity-controlled incubator with 5% CO2. Antibiotics were usedroutinely, including penicillin (100 IU/ml) and strepto-mycin (100 mg/ml).

VirusesCTL102 is an E1-deleted, replication-defective adeno-virus that expresses NTR from the CMV promoter,23

and was kindly provided by ML Laboratories plc.,Keele, Staffordshire, UK. The structure of CRAd-NTR(PS1217H6) is shown in Figure 1. Like CTL102, itis based on our earlier E1, E3-deleted Ad5 vectors;20,23,45

sequences between the SacII site at bp 358 and 2321containing E1 and the E1B-19K gene were derived fromdl1520, except that the XbaI site within E1A was silentlymutated to facilitate later cloning steps. This wasseparated from the CMV-NTR expression cassette by a296 bp fragment containing the 30 end of the bovinegrowth hormone gene, derived from pcDNA3 (Invitro-gen). Viruses were generated from plasmid DNA,propagated and plaque-titrated using 911 cells. Virusstocks were purified by caesium chloride densitygradient centrifugation.

Quantitating virus replicationReplication of adenovirus DNA in vitro was monitoredby dot blotting. Cells were infected with CRAd-NTR(PS1217H6) or CTL102 at a dose of 0.5 PFU/cell.At various times up to 5 days, DNA extraction wasperformed by the method of Hirt.46 The DNA sampleswere transferred to a positively charged nylon mem-brane and detected by hybridization with a radioactivea-32P-dCTP-labelled DNA probe, containing bp 31 994–34 930 of Ad5-WT. Image collection and quantitationwere performed by exposing the blot for 2 h to a storagephosphor screen, which was then read using a Typhoon8600 Variable Mode Imager (Molecular Dynamics, HighWycombe, UK). Calibration was performed using aplasmid standard.

To monitor replication of infectious virus particles invitro, SW480 or WiDr cells (1�105 cells per 6 cm dish)

were infected with CRAd-NTR(PS1217H6) or CTL102(10 PFU/cell) and harvested at intervals over 4–5 days.Virus particles were released by three rounds of freezingand thawing, and the titre of infectious viral particleswas determined by plaque assay on 911 cells.

Quantitative polymerase chain reaction (Q-PCR) wasperformed to quantitate adenoviral DNA in freshlyresected tumours. The oligonucleotide sequences of theDNA probe, forward and reverse primers were withinthe adenovirus type 5 fibre gene. Primers and probewithin E1A were used to monitor stocks of the E1-deletedCTL102 virus for contamination. Samples were analysedusing an ABI Prisms 7700 Sequence Detector andsoftware (Applied Biosystems).

Analysis of NTR expressionProtein samples were prepared from cells cultured in6 cm dishes by washing with PBS, then harvesting in300 ml of lysis buffer (20 mM Tris-HCl pH 7.6, 250 mM

NaCl, 3 mM EGTA, 3 mM EDTA, 0.5% Triton X-100,10 mg/ml each of aprotinin and leupeptin (Sigma)).Protein concentrations were determined using Bio-Radprotein assay solution (BioRad Laboratories, Munchen,Germany).

Protein and DNA samples were prepared fromtumour tissue using TRIzol sReagent (Invitrogen)according to the manufacturer’s instruction.

To detect NTR expression, 50 mg (from cultured cells)or 100 mg (from tumours) protein sample per well wasfractionated by SDS–PAGE and transferred to nitrocellu-lose membranes. NTR was detected using a polyclonalsheep anti-NTR antibody (kindly provided by MLLaboratories plc., Keele, Staffordshire, UK) diluted1:20 000 and a donkey anti-sheep antibody conjugatedwith horseradish peroxidase (Sigma) diluted 1:2000. As acontrol, actin was detected using mouse anti-human a-actin antibody diluted 1:5000 (Sigma), followed by aperoxidase-conjugated goat anti-mouse antibody diluted1:1000 (DAKO).

Combination cytotoxicity assayA total of 5000 cells per well were plated in a 96-wellplate. The next day, cells were counted and infected withCRAd-NTR(PS1217H6) or CTL102, or mock-infected for2 h. At 48 h after infection, cells were exposed to variousconcentrations of CB1954 (kindly provided by MLLaboratories plc., Keele, Staffordshire, UK) for 4 days,before determination of cell viability using the MTScytotoxicity assay (CellTiter 96s Aqueous One SolutionCell Proliferation Assay kit, Promega, Madison, WI,USA).

In situ analysis of NTR expression and virus replicationin colorectal cancer xenograftsAll experimental procedures using animals were per-formed under the authority of UK Home Office Projectand Personal licences, and with institutional ethicalcommittee approval. Tumours were seeded in immuno-deficient Balb/C nude mice aged 6–8 weeks, by injecting5� 106 SW480 colon carcinoma cells subcutaneously intothe flank. When the tumours reached 5–8 mm diameter,they were injected with CRAd-NTR(PS1217H6) orCTL102 (1�108 PFU in 40 ml) by single injection, thenresected at selected time points, fixed in formalin,

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1134

Gene Therapy

paraffin-embedded and sectioned at 3 mm thickness. Theimmunostaining for NTR protein was as describedpreviously,22 with the addition of an initial antigenretrieval step involving microwaving the samples at800 W in 1 l citrate buffer (pH 5.8) for 20 min. NTR wasdetected using polyclonal sheep anti-NTR antibody(diluted 1:2000 in PBS) at room temperature for 1 h andwith a secondary, rabbit anti-sheep antibody (VectorLaboratories, Burlingame, CA, USA) diluted 1:200 inPBS, at room temperature for 30 min. This was followedby incubation with ABC mixture and AEC mixture todevelop colour (Vector Laboratories). Finally, counter-staining was performed using Mayer’s haematoxylindye. Adenovirus capsid proteins were detected usingrabbit anti-Ad5 antibody.47

For detection of viral DNA by Q-PCR, or NTR byWestern blotting, the tumours were grown as above butinjected once with 2� 108 PFU of the viruses.

Therapeutic colorectal cancer modelBalb/C nude mice were seeded with SW480 tumours asabove. When the tumours reached 5–8 mm diameter, themice were assigned at random to receive intratumoralinjection of CRAd-NTR(PS1217H6), CTL102 or vehicle(PBS). Tumours were injected with virus or PBS on 5consecutive days, 2� 108 PFU per 40 ml injection witheach injection in a different direction in order tomaximize the distribution of virus within the tumours.At 48 h after the last virus injection, half the animals ineach group were assigned to receive either PBS (200 ml/day) or CB1954 (25 mg/kg/day) for 4 consecutive days,by intraperitoneal injection. Each of the treatment groupscontained seven mice. Tumour size was measured twiceper week with callipers and the volume was estimatedby the formula: volume¼width2� length/2.

Acknowledgements

This work was supported by the UK Medical ResearchCouncil Grant G9806623. MJC acknowledges the supportof Mackay Memorial Hospital, Taipei, Taiwan. We thankML Laboratories plc. for CTL102, CB1954 and sheep anti-NTR antibody, and D Onion for Q-PCR analysis ofCTL102 virus stocks.

References

1 Vile R, Ando D, Kirn D. The oncolytic virotherapy treatmentplatform for cancer: unique biological and biosafety points toconsider. Cancer Gene Ther 2002; 9: 1062–1067.

2 Barker DD, Berk AJ. Adenovirus proteins from both E1B readingframes are required for transformation of rodent cells by viralinfection and DNA transfection. Virology 1987; 156: 107–121.

3 Bischoff JR et al. An adenovirus mutant that replicates selectivelyin p53-deficient human tumor-cells. Science 1996; 274: 373–376.

4 Ries SJ et al. Loss of p14ARF in tumor cells facilitates replicationof the adenovirus mutant dl1520 (ONYX-015). Nat Med 2000; 6:1128–1133.

5 Kirn D. Clinical research results with dl1520 (Onyx-015), areplication-selective adenovirus for the treatment of cancer:what have we learned? Gene Therapy 2001; 8: 89–98.

6 Khuri FR et al. A controlled trial of intratumoral ONYX-015, aselectively-replicating adenovirus, in combination with cisplatin

and 5-fluorouracil in patients with recurrent head and neckcancer. Nat Med 2000; 6: 879–885.

7 Reid T et al. Hepatic arterial infusion of a replication-selectiveoncolytic adenovirus (dl1520): phase II viral, immunologic, andclinical endpoints. Cancer Res 2002; 62: 6070–6079.

8 Huber BE et al. In vivo antitumor activity of 5-fluorocytosine onhuman colorectal carcinoma cells genetically modified to expresscytosine deaminase. Cancer Res 1993; 53: 4619–4626.

9 Culver KW et al. Gene therapy for the treatment of malignantbrain tumors with in vivo tumor transduction with the herpessimplex thymidine kinase gene ganciclovir system. Hum GeneTher 1994; 5: 343–379.

10 Mullen CA, Kilstrup M, Blaese RM. Transfer of the bacterial genefor cytosine deaminase to mammalian cells confers lethalsensitivity to 5-fluorocytosine: a negative selection system. ProcNatl Acad Sci USA 1992; 89: 33–37.

11 Hamstra DA et al. Enzyme/prodrug therapy for head and neckcancer using a catalytically superior cytosine deaminase. HumGene Ther 1999; 10: 1993–2003.

12 Bridgewater JA et al. Expression of the bacterial nitroreductaseenzyme in mammalian cells renders them selectively sensitiveto killing by the prodrug CB1954. Eur J Cancer 1995; 31A:2362–2370.

13 Green NK et al. Sensitization of colorectal and pancreatic cancercell lines to the prodrug 5-(aziridin-1-yl)-2,4-dinitrobenzamide(CB1954) by retroviral transduction and expression of the E. colinitroreductase gene. Cancer Gene Ther 1997; 4: 229–238.

14 Grove JI et al. Virus directed enzyme prodrug therapy usingCB1954. Anti-Cancer Drug Des 1999; 14: 461–472.

15 Wildner O, Blaese RM, Morris JC. Therapy of colon cancer withoncolytic adenovirus is enhanced by the addition of herpessimplex virus-thymidine kinase. Cancer Res 1999; 59: 410–413.

16 Wildner O et al. Adenoviral vectors capable of replicationimprove the efficacy of HSVtk/GCV suicide gene therapy ofcancer. Gene Therapy 1999; 6: 57–62.

17 Freytag SO et al. A novel three-pronged approach to kill cancercells selectively: concomitant viral, double suicide gene, andradiotherapy. Hum Gene Ther 1998; 9: 1323–1333.

18 Rogulski KR et al. Double suicide gene therapy augments theantitumor activity of a replication-competent lytic adenovirusthrough enhanced cytotoxicity and radiosensitization. Hum GeneTher 2000; 11: 67–76.

19 Knox RJ, Friedlos F, Boland M. The bioactivation of CB1954 andits use as a prodrug in antibody-directed enzyme prodrugtherapy (ADEPT). Cancer Metast Rev 1993; 12: 195–212.

20 Weedon SJ et al. Sensitisation of human carcinoma cells to theprodrug CB1954 by adenovirus vector-mediated expression ofE. coli nitroreductase. Int J Cancer 2000; 86: 848–854.

21 McNeish IA et al. Virus directed enzyme prodrug therapy forovarian and pancreatic cancer using retrovirally delivered E. colinitroreductase and CB1954. Gene Therapy 1998; 5: 1061–1069.

22 Djeha AH et al. Expression of E. coli B nitroreductase inestablished human tumour xenografts in mice results in potentanti-tumour and bystander effects upon systemic administrationof the prodrug CB1954. Cancer Gene Ther 2000; 7: 721–731.

23 Djeha AH et al. Combined adenovirus-mediated nitroreductasegene delivery and CB1954 treatment: a well-tolerated therapy forestablished solid tumors. Mol Ther 2001; 3: 233–240.

24 Palmer D et al. Virus directed enzyme prodrug therapy:intratumoral administration of a replication-deficient adeno-virus encoding nitroreductase to patients with resectable livercancer. J Clin Oncol 2004; 22: (Published online 1st May 2004).

25 Steinwaerder DS, Carlson CA, Lieber A. DNA replication offirst-generation adenovirus vectors in tumor cells. Hum GeneTher 2000; 11: 1933–1948.

26 Heise CC, Williams A, Olesch J, Kirn DH. Efficacy of areplication-competent adenovirus (ONYX-015) following

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1135

Gene Therapy

intratumoral injection: intratumoral spread and distributioneffects. Cancer Gene Ther 1999; 6: 499–504.

27 Chung-Faye G et al. Virus-directed, enzyme prodrug therapywith nitroimidazole reductase: a phase I and pharmacokineticstudy of its prodrug, CB1954. Clin Cancer Res 2001; 7: 2662–2668.

28 Searle PF, Spiers I, Simpson J, James ND. Cancer gene therapy:from science to clinical trials. Drug Deliv Systems Sci 2002; 2:5–13.

29 Sausville EA. Genes in the service of therapeutic index: progressfor virus-directed enzyme prodrug therapy. J Clin Onc 2004; 22:1535–1537.

30 Turnell AS, Grand RJ, Gallimore PH. The replicative capacities oflarge E1B-null group A and group C adenoviruses areindependent of host cell p53 status. J Virol 1999; 73: 2074–2083.

31 Edwards SJ et al. Evidence that replication of the antitumoradenovirus ONYX-015 is not controlled by the p53 and p14ARFtumor suppressor genes. J Virol 2002; 76: 12483–12490.

32 Boland MP, Knox RJ, Roberts JJ. The differences in kinetics of ratand human DT diaphorase result in a differential sensitivity ofderived cell lines to CB1954 (5-(aziridin-1-yl)-2,4-dinitrobenzamide). Biochem Pharmacol 1991; 41: 867–875.

33 Green NK et al. Immune enhancement of nitroreductase-inducedcytotoxicity: studies using a bicistronic adenovirus vector. Int JCancer 2003; 104: 104–112.

34 Hawkins LK, Hermiston T. Gene delivery from the E3 region ofreplicating human adenovirus: evaluation of the E3B region.Gene Therapy 2001; 8: 1142–1148.

35 Nanda D et al. Treatment of malignant gliomas with a replicatingadenoviral vector expressing herpes simplex virus-thymidinekinase. Cancer Res 2001; 61: 8743–8750.

36 Babiss LE, Ginsberg HS, Darnell Jr JE. Adenovirus E1B proteinsare required for accumulation of late viral mRNA and for effectson cellular mRNA translation and transport. Mol Cell Biol 1985;5: 2552–2558.

37 Wildner O, Morris JC. The role of the E1B 55 kDa gene productin oncolytic adenoviral vectors expressing herpes simplex virus-

tk: assessment of antitumor efficacy and toxicity. Cancer Res 2000;60: 4167–4174.

38 Heise C et al. An adenovirus E1A mutant that demonstratespotent and selective systemic anti-tumoral efficacy. Nat Med2000; 6: 1134–1139.

39 Nettelbeck DM et al. Novel oncolytic adenoviruses targeted tomelanoma: specific viral replication and cytolysis by expressionof E1A mutants from the tyrosinase enhancer/promoter. CancerRes 2002; 62: 4663–4670.

40 Coukos G et al. Oncolytic herpes simplex virus-1 lacking ICP34.5induces p53-independent death and is efficacious againstchemotherapy-resistant ovarian cancer. Clin Cancer Res 2000; 6:3342–3353.

41 Rodriguez R et al. Prostate attenuated replication competentadenovirus (ARCA) CN706: a selective cytotoxic for prostate-specific antigen-positive prostate cancer cells. Cancer Res 1997;57: 2559–2563.

42 Gallimore PH et al. Adenovirus type 12 early region 1B 54Kprotein significantly extends the life span of normal mammaliancells in culture. J Virol 1997; 71: 6629–6640.

43 Fallaux FJ et al. Characterization of 911: a new helper cell line forthe titration and propagation of early-region-1-deletedadenoviral vectors. Hum Gene Ther 1996; 7: 215–222.

44 Berry IJ, Burns JE, Parkinson EK. Assignment of two humanepidermal squamous cell carcinomas cell lines to more than onecomplementation group for the immortal phenotype. MolCarcinogen 1994; 9: 134–142.

45 Latham JP, Searle PF, Mautner V, James ND. Prostate-specificantigen promoter/enhancer driven gene therapy for prostatecancer: construction and testing of a tissue-specific adenovirusvector. Cancer Res 2000; 60: 334–341.

46 Hirt B. Selective extraction of polyoma DNA from infectedmouse cells. J Mol Biol 1967; 26: 365–369.

47 Stallwood Y, Fisher KD, Gallimore PH, Mautner V.Neutralisation of adenovirus infectivity by ascitic fluid fromovarian cancer patients. Gene Therapy 2000; 7: 637–643.

Oncolytic adenovirus expressing nitroreductaseM-J Chen et al

1136

Gene Therapy