research article genetic prodrug activation therapy (gpat) in...

Gene Therapy (2001) 8, 557–567 2001 Nature Publishing Group All rights reserved 0969-7128/01 $15.00

www.nature.com/gt

RESEARCH ARTICLE

Genetic prodrug activation therapy (GPAT) in two ratprostate models generates an immune bystandereffect and can be monitored by magnetic resonancetechniques

JD Eaton1,2, MJA Perry1,2, SM Todryk1, RA Mazucco3, RS Kirby2, JR Griffiths3 and AG Dalgleish11Division of Oncology and 3Department of Biochemistry and Immunology, St George’s Hospital Medical School, and 2Department ofUrology, St George’s Hospital, London, UK

Treatment of hormone refractory prostate cancer requires newtreatment strategies. Genetic prodrug activation therapy(GPAT) may provide a new therapeutic avenue. In this studythe antitumour efficacy of the gene encoding herpes simplexvirus thymidine kinase (HSVtk) activating the prodrug gan-ciclovir (GCV) was compared in two models of ectopic(subcutaneous) rat prostate cancer. Both models, which differin their characteristics, were previously shown to be weaklyimmunogenic but susceptible to immunotherapy. Tumour celllines were stably transfected with HSVtk and were renderedhighly sensitive to GCV. Little or no bystander killing effectwas observed by tk-transfected cells on wild-type cells in vitro.However, a significant in vivo bystander effect was observedsuggesting an immune-mediated response. Indeed, such animmune response was capable of slowing the growth of dis-tant wild-type tumours and increased overall animal survival.A T helper 1 immune response was generated as a result ofGCV activation and cell kill, demonstrated by the secretion ofIFNg by cultured splenocytes in response to tumour cells.

Keywords: prostate cancer; suicide gene therapy; GPAT; immunity

IntroductionHormone refractory prostate cancer currently has a poorprognosis, as once a patient has progressed to this stageno treatment has been shown to alter overall survival.Increased testing of serum prostate-specific antigen (PSA)allows for earlier diagnosis of malignancy but curativesuccess of radical prostatectomy or radiotherapy is lim-ited to less than two-thirds of patients. Moreover, meta-static disease can be treated with anti-androgen therapybut overall survival from the time of diagnosis is between2 and 3 years. In 1996, within the UK, 9700 men diedfrom prostate cancer (Office of National Statistics, 1997;www.statistics.gov.uk). There is therefore a demand fornovel therapies which may potentially impact upon hor-mone refractory disease. Among the many treatment

Correspondence: AG Dalgleish, Division of Oncology, St George’s Hospi-tal Medical School, Cranmer Terrace, London SW17 0RE, UKThe first two authors contributed equally to this studyReceived 30 May 2000; accepted 15 January 2001

BrDU staining of tk-transfected cells treated with GCV in vitrosuggested apoptotic cell death, but Annexin V staining wasless marked for one of the cell lines. Serial in vivo monitoringby non-invasive magnetic resonance spectroscopy (MRS) ofthe tk-transfected MATLyLu tumours demonstrated adecreased ATP/Pi ratio (a measure of cell energy status) dur-ing growth and an increase in the ATP/Pi ratio duringregression initiated by treatment with GCV. Further, significantdifferences were found in the phosphomonester (PME) to totalphosphate (SP) ratios in treated compared with untreatedtumours, a result rarely seen in animal models, but commonlyobserved in patients. This study showed that a Th1-biasedimmune response generated by killing prostate tumour cellswith tk/GCV can kill distant as well as local wild-type tumourcells. These findings suggest that GPAT may have a potentialapplication in patients with both confined and metastatic pros-tate cancer and MRS may provide a method of monitoringresponse to treatment. Gene Therapy (2001) 8, 557–567.

strategies currently being considered, gene therapyappears to have potential to make an impact on prostatecancer.1 The area of cancer gene therapy overlaps con-siderably with immunotherapy in that many of the genetransfer strategies involve transduction with genes enco-ding directly or indirectly immunostimulatory molecules.Genes can be transferred either directly in situ (in vivo)or to isolated tumour cells (ex vivo) which are thenirradiated and returned to the patient as a ‘cancer vac-cine’.2 Since prostate cancer cells express specific anti-gens, this cancer appears to be a potential target forimmune responses. Early clinical trials using wholetumour cells transduced with granulocyte–macrophagecolony-stimulating factor (GM-CSF),3 or dendritic cellspulsed with prostate-antigen peptides4 resulted inpatients demonstrating tumour-specific immuneresponses. This suggests that prostate cancer may beamenable to immune-based therapy even though clinicalresponses in these trials were limited due to theadvanced stage of disease in the patients. In addition, thepresence of tumour-infiltrating lymphocytes (TIL) within

GPAT in prostate cancerJD Eaton et al

558

Gene Therapy

prostate tumours has been correlated with a better prog-nosis,5 further suggesting the role for immune manipu-lation as a treatment strategy. Even nonspecific immuno-therapy with M. vaccae has been shown to have potentialantitumour activity.6 This may be countering the sup-pression of cell-mediated immunity which occurs inmany malignancies such as melanoma,7 colorectal can-cer,8 as well as prostate cancer. It is known that manypatients undergoing radical prostatectomy are left withminimal residual disease and thus may be in a situationwhere gene therapy/immunotherapy has good efficacyagainst the small residual tumour burden. However,there are specific characteristics of prostate cancer thatmay result in immune evasion, such as loss of expressionof MHC class I molecules,9 disruption of other antigen-presenting machinery, and the expression of immuno-suppressive molecules such as TGFb, IL-10 and Fasligand.

Genetic prodrug activation therapy (GPAT) is aninnovative approach that can kill tumour cells byinserting ‘suicide’ genes into cancer cells, a strategy alsoreferred to as suicide gene therapy. The most frequentlydescribed GPAT strategy is the herpes simplex virus thy-midine kinase (HSVtk) enzyme/ganciclovir (GCV) pro-drug system. HSVtk phosphorylates GCV and this pro-duct incorporates into DNA during cell division, thuskilling the cell. HSVtk/GCV is a potential avenue for tre-ating cancer by the direct transduction and killing of aproportion of tumour cells in vivo and by the associatedlocal bystander killing effect.10 In addition, the immuneresponses generated by this process are important for thelocal bystander effect and may also affect distant metast-ases.11,12 GPAT using the HSVtk system has alreadyentered clinical trials for ovarian cancer13 and mesotheli-oma14 using adenoviral delivery of HSVtk. In particular,a recent trial of HSVtk/GCV for organ-confined prostatecancer post-irradiation has shown some evidence of anti-tumour activity using a falling serum PSA as a surrogatemarker of response.15 However, methods for direct moni-toring of a therapeutic effect are needed, especially iftumours no longer secrete PSA.

Modelling prostate cancer in animals is required inpre-clinical ‘proof-of-principle’ studies and here wedescribe two rat models of prostate cancer which showsome characteristics of the human disease. We show thatthe magnitude of the bystander effect associated withHSVtk/GCV is more significant in vivo than in vitro andthat the associated immunity that is elicited in vivocan effect responses in distant tumours that representmetastases.16

Magnetic resonance spectroscopy (MRS) has theunique ability to measure the concentrations of chemicalsin living tissue noninvasively.17 The proteins expressedas a result of gene transfection may alter the concen-tration of cellular metabolites, thus the measurement oftumour biochemistry by MRS may represent an excellentmethod for monitoring GPAT strategies for cancer. Sincethe first in vivo MR spectra were acquired in tumours inrats18 and then in patients,19 the links between the bio-chemical characteristics of tumours and MRS have beenwell documented. A large number of studies have showncorrelations between different cancer treatments intumour models and patients and changes in MR spec-tra.20 A number of reports have also demonstrated theeffectiveness of MRS and magnetic resonance imaging

(MRI) for monitoring HSVtk gene therapy treatments.MRI and MRS have been particularly useful in animalmodels in otherwise inaccessible tumours such as inbrain.21–24 Lately, Hakumäki et al25 have demonstratedthat GCV-induced apoptosis in tk-transfected experi-mental gliomas could be detected by 1H MRS. In thestudy reported here, we demonstrate that the efficacy ofthe treatment can be monitored with in vivo MRIand 31P MRS.

Results

In vitro characteristics of cellsFollowing retroviral transfection of the tumour cells, neo-mycin-resistant clones were picked, replated and propa-gated. Clones that showed the greatest susceptibility tokilling with GCV were selected for further study. Theseclones were totally eradicated with 5 mg/ml GCV, whilstthe growth of the wild-type cells was not affected at thisconcentration. The increase in sensitivity to GCV of thetk-transfected clones was compared to wild-type cellsover a wide range of GCV concentrations (0.001 to100 mg/ml) (Figure 1 a and b). Thus the LC50 of MLL wasdecreased from 9 to 0.006 mg/ml whilst the LC50 of PAIIIwas decreased from 20 to 0.08 mg/ml by tk-transfection.

In order to test for a possible bystander killing effect,wild-type (WT) and tk-transfected (tk) tumour cells weremixed in the following proportions; 100% WT/0% tk,90% WT/10% tk, 70% WT/30% tk, 50% WT/50% tk, 30%WT/70% tk, 10% WT/90% tk, 0% WT/100% tk, andtreated with 5 mg/ml GCV. At either time point of 5 or7 days no significant in vitro bystander effect wasobserved for either MLL or PAIII cell lines (Figure 2 aand b). Indeed, the rapid growth of MLL was very evi-dent in these experiments where the cells simplyexpanded to fill the culture wells.

In vivo experimentsThe in vivo growth of the tk-transfected tumour line wasfound to be slightly retarded compared to the wild-typeline (Figure 3a) in the MLL model. This difference wasnot found to be statistically significant (P = 0.09). In thePAIII model the growth rates of wild-type and tk-trans-fected tumours were very similar (Figure 3b). In order totest the in vivo sensitivity of the tumours to GCV, aninoculum of 1 × 105 MLL tumour cells and 1 × 106 PAIIItumour cells was injected subcutaneously into syngeneicrats. After tumours had reached 5 mm in diameter therats were given GCV i.p. daily for 5 days at the followingdoses: 10, 50 and 100 mg/kg. All these doses eradicatedthe tk tumours entirely (Figure 3c and d). Animals inwhich the tumours had receded were monitored further.In three out of eight of the Cops and four out of eight ofthe Lobs that had received 10 mg/kg of GCV thetumours grew back. This was not evident in the animalsreceiving higher doses of GCV. In addition, the weightof the animals during treatment with the various dosesof GCV was monitored. Animals receiving 100 mg/kgshowed the largest fluctuation in weight with an averageweight loss of 20% body weight during the 5 day treat-ment course. Thus the dose of 50 mg/kg was deemed tobe the optimal dose in terms of efficacy and toxicity.

In view of the lack of demonstrated bystander killingin vitro, an in vivo bystander effect was examined. Mix-


559a

b

Figure 1 In vitro sensitivity of wild-type and tk-transfected MLL andPAIII cells to GCV. Wild-type and tk-transfected cells were plated out in24-well plates and medium was added containing a wide range of concen-trations of GCV (0.001 to 100 mg/ml). Cell viability was assessed after 3days using an MTT assay. The results are shown as percentage cell sur-vival of both cell types at the range of GCV concentrations. The LC50 ofMLL (a) was decreased from 9 to 0.006 mg/ml by tk-transfection, whilethe LC50 of PAII (b) was decreased from 20 to 0.08 mg/ml.

tures of wild-type and tk-transfected lines in the follow-ing proportions; 100% WT/0% tk, 75% WT/25% tk, 50%WT/50% tk, 25% WT/75% tk, 0% WT/100% tk, wereinjected subcutaneously as in previous experiments,allowed to grow to 5 mm and then treated with50 mg/kg GCV i.p. Significant bystander effect occurredin vivo. At day 20, 25% Mtk resulted in a 77.6% reductionin tumour size over wild-type alone (P = 0.007) and atday 26, 50% Ptk caused a 57.9% reduction in their size(P = 0.03) (Figure 4a and b). Since there was no in vitrobystander effect observed this suggested that there wasan immune component to this effect. Rats that had com-pletely rejected their initial tumours were rechallenged

Gene Therapy

a

b

Figure 2 In vitro bystander effect. Wild-type (WT) and tk-transfected (tk)MLL (a) or PAIII (b) tumour cells were mixed in the following pro-portions; 100% WT/0% tk, 90% WT/10% tk, 70% WT/30% tk, 50%WT/50% tk, 30% WT/70% tk, 10% WT/90% tk, 0% WT/100% tk, wereplated out in 24-well plates. After 24 h the medium was changed formedium containing 5 mg/ml GCV. Cell viability was assessed after 5 daysusing an MTT assay. No significant bystander effect was observed.

with wild-type cells, a significant delay in tumour devel-opment was observed with 12.5% of Cops and 50% ofLobs totally protected at 28 days and 55 days, respect-ively, after rechallenge (Figure 4c and d).

We next investigated whether tk-killing could have aneffect on established wild-type tumours injected andgrown at a site distant from the site where tk-transfectedtumours were established in the same animal. Tumours


560

Gene Therapy

a c

b d

Figure 3 In vivo growth of tk-transfected tumour cells. An inoculum of 1 × 105 MLL/Mtk, or 1 × 106 PAIII/Ptk tumour cells was injected subcutaneouslyinto the shaven thigh of Cop or Lob rats respectively (n = 8 per group). The growth of the Mtk tumour line was found to be slightly retarded comparedwith the wild-type line (a). This difference was not statistically significant (P = 0.09). Growth rates of wild-type and Ptk tumours were very similar(b). An inoculum of 1 × 105 Mtk (c) or 1 × 106 Ptk (d) tumour cells was injected subcutaneously into the shaven thigh of Cop or Lob rats, respectively.After tumours had reached 5 mm in diameter the rats were treated with GCV i.p. daily for 5 days at the following doses: 10, 50 and 100 mg/kg. Acontrol group received PBS i.p. for the same period of time (n = 8 per group). All doses of GCV eradicated the tk-transfected tumours.

were grown bilaterally on the opposite flanks of the ani-mals with two groups of animals receiving tk-transfectedtumours on the left thigh and wild-type tumours on theright, a final group received bilateral wild-type tumours.Wild-type tumours grew at a normal rate when theywere grown on both flanks of the rats with or without

GCV treatment (Figure 5a and b). Growth of tk-trans-fected tumours was abrogated by GCV and this killingalso significantly slowed the growth of wild-typetumours on the opposite flank. This effect on wild-typetumours was associated with a significant increase in sur-vival compared with the animals with bilateral wild-type


561ca

db

Figure 4 In vivo bystander effect and generation of immune protection. An inoculum of 1 × 105 MLL/Mtk, or 1 × 106 PAIII/Ptk tumour cells, consistingof wild-type and tk-transfected cells in the following proportions; 100% WT/0% tk, 75% WT/25% tk, 50% WT/50% tk, 25% WT/75% tk, 0% WT/100%tk, were injected subcutaneously into the shaven thigh of Cop (a) and Lob (b) rats, respectively (n = 10 per group). After tumours had reached 5 mmin diameter the rats were treated with 50 mg/kg GCV i.p. daily for 5 days. Significant bystander effect occurred in vivo. At day 20, 25% Mtk resultedin a 77.6% reduction in tumour size over wild-type alone (P = 0.007) (a). At day 26, 50% Ptk resulted in a 57.9% reduction in tumour size over wild-type alone (P = 0.03) (b). In vivo rechallenge following tk tumour eradication. An inoculum of 1 × 105 Mtk, or 1 × 106 Ptk tumour cells were injectedinto the shaven left thigh of Cop (c) or Lob (d) rats, respectively (n = 8 per group). After 10 days the animals were treated with 50 mg/kg GCV i.p.daily for 5 days. All animals were followed up for a 2-week period following tumour eradication, to ensure no tumour recurrence. They were thenrechallenged with an inoculum of 1 × 105 MLL or 19 × 106 PAIII tumour cells to the right thigh. Control groups of naı̈ve animals received an identicalcell challenge (n = 6 per group). A significant delay in tumour development was observed in both the rechallenged groups compared with the nävegroups (P = 0.0043) (c), (P = 0.0001) (d).

Gene Therapy


562

Gene Therapy

a

b

Figure 5 Distant bystander effect in vivo. Three groups of Cop and Lobrats were used (n = 10 per group). Groups 1 and 2 received an inoculumof 1 × 105 Mtk or 1 × 106 Ptk cells to the left thigh and 1 × 105 MLL or1 × 106 PAIII cells to the right thigh. Group 3 received a bilateral inocu-lum of 1 × 105 MLL or 1 × 106 PAIII cells. Ten days after the initialtumour challenge, groups 1 and 3 received 50 mg/kg GCV i.p. for 5 days,group 2 received PBS i.p. for 5 days. In group 1 growth of the tk-trans-fected tumours was abrogated by GCV as well as the wild-type tumouron the opposite flank. At day 18 in the MLL/Mtk experiment this resultedin a significant difference (P = 0.02) between the volume of the wild-typetumours in group 1 compared with groups 2 and 3 (a). Likewise at day32 in the PAIII/Ptk experiment this difference was also significant(P = 0.01) (b). In group 2 the growth of the tk-transfected tumours wasnot affected by the PBS.

tumours treated with GCV and also the animals with tk-transfected and wild-type tumours treated with PBS(MLL/Mtk, P = 0.002; PAIII/Ptk, P = 0.004).

Immune responsesIn order to investigate the nature of the anti-tumourimmune responses following tk/GCV killing of tumourcells, spleens were taken from rats with established tk-transfected tumours treated with GCV and comparedwith a group of rats that received no GCV. Single celllymphocyte suspensions were prepared and incubated

Table 1 IFNg secretion (from rat splenocytes)

Treatment groups

Mtk + GCV Mtk Naı̈ve

Animal Nos1 .1400 289 1432 338 93 1623 .1400 313 290

Mtk, tk-transfect MAT Ly Lu.Rat splenocytes from treated, untreated or naı̈ve rats were culturedwith their tumour cells in vitro for 5 days. IFNg secretion was meas-ured by ELISA.

for 5 days with tumour cells. After this time the lympho-cytes were harvested and tested for CTL activity againstthe syngeneic cell line whilst the supernatants fromsplenic cultures were tested for secreted cytokines, IFNgand IL-4 by ELISA. On average, higher levels of IFNgwere detected in the group whose Mtk tumours hadreceived GCV treatment compared with those not receiv-ing GCV and naı̈ve controls (Table 1). However, no CTLactivity was detectable with the Cop/Mtk system. In theLob/Ptk system low levels of CTL activity were detectedin the GCV-treated group, which were significantlyhigher than those of the naı̈ve rats (Figure 6a). Tk-trans-fected tumours were removed from dead rats after GCVtreatment or PBS treatment, disaggregated, and stainedfor FACS analysis. All the tumours appeared to have aninfiltration of CD45+ leukocytes (data not shown).Tk/GCV tumour cell killing was associated with higherCD8+ infiltration, whilst equivalent CD4+ and NK infil-tration was observed in both groups.

ApoptosisA recent report has demonstrated that the mechanism oftumour cell death with tk/GCV may affect immuno-

Figure 6 CTL activity following tk/GCV treatment. CTL activity ofsplenocytes from Lob rats (naı̈ve or following tk/GCV eradication oftumour) were assayed by 51Cr release from labelled PAIII cells. Treatedrats showed significantly more activity. Mean +/− s.d. is given for threerats per group.


563genicity.26 To investigate this, Mtk and Ptk cells weretreated with 5 mg/ml GCV and then analysed by BrDUand Annexin V staining by FACS (Figure 7). Both Mtkand Ptk showed .70% BrDU staining following GCVtreatment. However, while Ptk showed considerableAnnexin V staining (66%) following GCV, Mtk showedonly 19.9%. In order to follow up these findings in termsof ‘danger signal’ generation, RNA was extracted fromGCV-treated and untreated cells. The RNA was reverse-transcribed and PCR was carried out to look for induciblerat hsp70 expression. No product was found in GCV-treated compared with heated (42°C) cells that expressedabundant hsp70 (data not shown).

Magnetic resonance imaging and spectroscopySerial 31P MR spectra from a representative Mtk tumourtreated with GCV are shown in Figure 8a. LikewiseFigure 8b shows images of a typical Mtk tumour shrink-ing and becoming walled off from the animal duringtreatment with GCV. Because of the inherent lack of sen-sitivity and relatively poor signal/noise of in vivo 31PMRS techniques only tumours treated with 0 and 100 mgdoses of GCV were monitored where there was the bestchance of seeing significant differences. The tumourswere monitored by MRS for 5 days starting on the 10thday after inoculation. After 3 days of treatment, a signifi-cant difference (P , 0.05) was evident in the ratio of theb-ATP peak to the inorganic phosphate (Pi) peak (b-ATP/Pi) between tumours that had received daily GCVtreatment and untreated controls. After 4 days of treat-ment, this difference was still significant (P , 0.01) andin addition, a significant difference (P , 0.01) was alsoseen in the ratio of the phosphomonoester (PME) peaksto the total phosphate value (PME/SP) (Figure 8c).

a b

Figure 7 Apoptosis of tumour cells following tk/GCV treatment. Apoptosis of tumour cells was assayed by both BrDU staining (a) and by AnnexinV staining (b). No GCV is in red whilst GCV is in white. Figures indicate percentage of positively stained cells (within M1 marker) withoutGCV/with GCV.

Gene Therapy

Discussion

The direct killing effect of HSVtk/GCV on tumours isitself highly desirable. However, current vector tech-nology may not allow sufficient numbers of cells to betransduced with HSVtk to allow the eradication of anentire tumour upon exposure to GCV. A bystander effectcauses the death of untransduced tumour cells locally bytransfer of metabolites and by the generation of animmune response. This immune response may also havean impact on tumour cells that have metastasised.Indeed, even if this approach only manages to slowtumour growth, the prospect of improved palliation can-not be ignored. The aim of this study was to establish amodel of GPAT for prostate cancer that could providevaluable information about the activation system, how itfunctions and how various effects can be monitored. Thedemonstration of systemic anti-tumour immunity by adistant bystander effect has important clinical impli-cations.

Two rat prostate tumour models were used since theyshow different characteristics from each other that maymirror human disease. PAIII in the Lobund-Wistar rats isa slow-growing tumour that may be more akin to earlyhuman disease. MLL in the Copenhagen rats, the Dunningmodel of prostate cancer, on the other hand is a fast-grow-ing, aggressive tumour that may have some resemblanceto hormone refractory disease. In terms of immuno-genicity we had previously demonstrated that the two celllines showed moderate immune protection when com-bined in whole cell vaccines with a mycobacterial adju-tant.27 More recently we have shown cross-reactivitybetween the lines when MLL was used as an allogeneicvaccine without adjuvant for protection against PAIII in


564

Gene Therapy

Figure 8 Serial in vivo 31P MRS during GCV treatment. An inoculum of 1 × 106 Mtk tumour cells were injected subcutaneously into the shaven thighof Cop rats. After 10 days tumour growth, rats were treated with GCV i.p. daily for 5 days at 100 mg/kg and serial in vivo 31P MR spectra acquiredfrom the Mtk tumours. PME, phosphomonoesters; Pi, inorganic phosphate; PDE, phosphodiesters; g, a and b-ATP, adenosine triphosphate; NADP,nicotinamide adenine dinucleotide phosphate. (b) Serial in vivo 1H MR imaging during GCV treatment. An inoculum of 1 × 106 Mtk tumour cells wasinjected subcutaneously into the shaven thigh of Cop rats. After 10 days tumour growth, rats were treated with GCV i.p. daily for 5 days at 100 mg/kgand serial in vivo 1H images acquired from each tumour. Images showed GCV-treated Mtk tumours rapidly regressing and becoming walled off fromthe animal. (c) Serial in vivo 31P MRS measurements of ATP/Pi and PME/SP ratios. An inoculum of 1 × 106 Mtk tumour cells was injected subcutane-ously into the shaven thigh of Cop rats. Rats were treated with GCV i.p. daily for 5 days at 100 mg/kg (n = 6) or saline (n = 8). Serial in vivo 31P MRmeasurements were made on each tumour and the ATP/Pi and PME/SP ratios determined. Tumours in GCV-treated rats displayed an increase inATP/Pi and decrease in PME/SP concomitant with treatment. Untreated tumour growth correlated with a marked decrease in ATP/Pi and no changein PME/SP. Points marked with an asterisk on each graph were significantly different from each other (P , 0.05, n = 5 or 6). SP was calculated fromeach spectrum as the sum of the areas under all metabolite peaks in the spectrum.


565Lobund-Wistar rats.28 Other groups have also shown thatMLL vaccines expressing IL-229 or GM-CSF30 generatedimmune protection. Thus the tumour models are suscep-tible to immune-based therapy. Transfection of HSVtk intoboth tumour lines imparted GCV sensitivity which causedtheir eradication in vitro and in vivo. Interestingly, no sig-nificant bystander effect on wild-type cells was observedin vitro. This contrasts to previous studies of killing ofprostate tumour cells where a murine tumour cell line anda human cell line (DU-145) did show in vitro bystanderkilling after transduction. However, a further human cellline (PC3) failed to show significant bystander effect usingthe same GPAT system.31

A significant local bystander effect was observed invivo, suggesting an immune basis for tumour destruction.Animals that were rechallenged with wild-type cells, fol-lowing previous eradication of tk-transfected tumours byGCV, showed significant delay in tumour growth andseveral animals were completely protected.

The effect of tk killing on established wild-typetumours distant from the tk-transfected tumour wasparticularly significant. Distant, non-contiguous wild-type tumours represent metastases similar to thoseoccurring in prostate cancer patients.

There was a significant reduction in tumour volumeand increased survival of animals as a result of the tk-killing. This study has shown that an anti-tumourimmune response can be raised after tk/GCV treatmentand that this may be beneficial to survival againstmetastatic disease.

The nature of the immune response was examined. Thesecretion of IFNg by splenocytes and CTL activity in vitro,and CD8+ cell infiltration of tumours in vivo was in keep-ing with a cell-mediated Th1 response, which is similarto previous observations in mouse tumour models.32

Recent evidence has suggested that the mechanism bywhich tumour cells die (apoptosis versus necrosis) withtk/GCV may influence the immune response generated.26Indeed, necrosis was considered more immunogenic thanapoptosis, which is the ‘physiological’ way in which cellsdie. Using BrDU staining both cell lines appeared to dieby apoptosis rather than necrosis. Annexin V staining, amarker of early apoptosis, suggested that apoptosisoccurred to a greater extent with Ptk than Mtk.Expression of the ‘danger signal’ hsp 70, which was pre-viously expressed in a mouse tumour model in associ-ation with necrosis was not found in either GCV-treatedcell line. This does not exclude the expression of otherdanger signals but suggests apoptosis is the dominantmode of cell death and facilitated ‘immune priming’.Immune priming with HSVtk/GCV has been furtherimproved in other studies by cotransfection of thetumour cells with tk and cytokine genes.33 Potentialimprovement in therapy could be effected by combi-nation with whole tumour cell vaccination.27 Indeed thetumours treated with 10 mg/kg of GCV that eventuallyregrew may be due to the need to induce a strong prim-ing immune response which our results suggest may bedose-related, hence further priming using such contri-butions as cytokines may further enhance effective anti-tumour immunity.

In a previous in vivo 31P MRS study, Stegman et al24showed that treatment of tk transfected C6 gliomas withGCV did not significantly change the in vivo ATP/Pi ratioor PME levels from pretreatment values. In contrast, our

Gene Therapy

study shows that GCV treatment in our tumour model(tk-transfected MLL tumours) causes significant changesin tumour ATP/Pi and in PME/SP. Thus, in this model,effects of GCV therapy of tk-transfected tumours can bemonitored by 31P MRS. Further, we have shown that thechanges concomitant with treatment and tumourregression are similar to those seen in other treatments,such as chemotherapy or radiotherapy. These changesinclude an increase in the high energy phosphate metab-olites following treatment.20,34 and a decrease inuntreated tumour growth. In MR studies on the distantbystander effect (which is the subject of another paper)similar spectra and responses were obtained in theuntreated and treated transfected tumours as in thisstudy, whereas the distant wild-type tumour, whosegrowth rate was significantly decreased by GCV treat-ment, showed no change in the ATP/Pi ratio.

A number of mechanisms to explain these findingshave been put forward34,35 that related to the vascularityand oxygenation of the responding, regressing tumour.

The fall in PME/SP seen in this model has rarely beenseen in animal tumour models, but it is the most commoneffect observed in human tumours in patients. A meta-analysis by Negendank et al36 who reviewed 61 patientsreceiving chemo- or radiotherapy found that 38/47 ofthose whose PME signal decreased eventually respondedto treatment. In contrast, only 1/14 nonrespondersshowed a PME decrease. PMEs are synthesized by theenzymatic activity of ethanolamine and choline kinaseswhich catalyze the first step of phospholipid biosynthesisin vivo. An increase in the PME peaks is thought to beassociated with an increase in cell membrane synthesis.37

This investigation did not address the actual process oftransfer of the tk genes to the tumours in vivo. Adenoviralvectors are currently the vehicles of choice for such trans-duction in animal models31 and in early human trials ofprostate cancer15, where delivery directly into the pros-tate is feasible. This next step is the subject of currentinvestigation in our laboratory.

This study has a number of implications for using sucha GPAT approach in the clinic. The finding that theimmune response generated by HSVtk/GCV-killing oftumour cells in vivo causes local and distal tumourregression is pivotal to the value of this approach.Reliable monitoring of the effects of this process is vitaland studies by MRI and MRS both at the site of primarytreatment and at distal, nontransfected tumours are ongo-ing and will be reported separately. Immune responsescould also be monitored following treatment both withsystemic lymphocytes and histologically within tumours.These findings are encouraging for embarking on morecomprehensive clinical trials of GPAT in prostate cancer.

Materials and methods

AnimalsPathogen-free 12-week-old male Copenhagen rats (Cops)were obtained from B&K (Hull, UK) and 12-week-oldmale Lobund-Wistar rats (Lobs) were obtained from Har-lan (USA). All procedures were carried out under mildanaesthesia and in accordance with the UK Home Officeand ethical guidelines.


566

Gene Therapy

Cell linesThe MAT-LyLu (MLL) subline of the Dunning rat pros-tate cancer was obtained from ECACC and was culturedin RPMI 1640 (Sigma, Poole, UK), 10% heat inactivatedfoetal calf serum (Sigma), 2 mm glutamine (Sigma), peni-cillin 100 mg/ml (Life Technologies, Paisley, UK) andstreptomycin 100 mg/ml (Life Technologies) in 5% CO2.The PAIII rat prostate cancer cell line was kindly pro-vided by Dr Morris Pollard (Lobund Laboratory, Indiana,USA) and was cultured in DMEM (Sigma), 10% heat inac-tivated foetal calf serum (Sigma), 2 mm glutamine(Sigma), penicillin 100 mg/ml, streptomycin 100 mg/ml,in 5% CO2. The cells were passaged with 0.05% trypsinand 0.02% EDTA (Sigma) and were free from myco-plasma as assessed by the gene probe method (Gen-Probe, San Diego, CA, USA).

Transfection and in vitro studiesThe gene encoding HSVtk was cloned into pBabeNeo andthis construct was transfected into the AM12 retroviralpackaging cell line, as described previously.11 This pack-aging cell line was kindly provided by Dr R Vile (MayoClinic, Rochester, MN, USA). Viral supernatant harvestedfrom these cells was used to transfect the rat prostatetumour lines in the presence of 4 mg/ml polybrene.Clones were grown in neomicin (G418) (150 mg/ml forMLL and 500 mg/ml for PAIII) and then tested for theirsensitivity to killing by GCV (Cymevene, Roche) at arange of concentrations compared to wild-type cells. Tk-transfectants of MLL and PAIII were referred to as Mtkand Ptk, respectively. In vitro growth characteristics ofthe transfected cell lines were carried out in 24-wellplates (Falcon, Becton-Dickinson). Bystander effect wasinvestigated by mixing wild-type and tk-transfected cellsin various proportions followed by GCV treatment. Pro-portions of surviving cells were assessed by a standardMTT assay, read on a spectrometer at optical density550 nm.

Tumour growth in vivoThe tumour cells were prepared for tumour challenge bydetachment with trypsin/EDTA, washed three times inPBS and resuspended at the specified density in PBS.Prostatic tumours were induced ectopically by the injec-tion of 1 × 106 PAIII cells in 100 ml PBS subcutaneouslyinto the shaven thigh of each Lobund-Wistar rat. In thecase of Copenhagen rats the tumorigenic dose of MLLcells was 1 × 105 cells, as optimised previously.27 Tumourgrowth was assessed every 2–3 days using microcalipersto measure perpendicular diameters. To eradicate the tk-transfected tumours, escalating GCV doses (10, 50 and100 mg/kg) were administered in 100 ml of PBS intraperi-toneally. Bystander effect was investigated by mixing tk-transfected and wild-type cells in various proportionsbefore inoculation, followed by GCV treatment. Ratswere killed when the tumour reached 25 mm in one ofthe diameters or when ulceration or bleeding occurred.Tumour volumes were calculated using the standard for-mula p/6 d1 (d2)2, where d2 is the smaller of the twodiameters.

Immune responsesSpleens were taken from dead rats that had previouslyhad tumours with or without GCV treatment. Thespleens were passed through 100 mm filters to obtain a

single-cell suspension and then red blood cells werelysed with 0.87% ammonium chloride. The cells werewashed and resuspended in RPMI 1640 supplementedwith foetal calf serum/glutamine/penicillin/streptomycin as above plus 50 mm 2-mercaptoethanol and10 U/ml human rIL-2 (recombinant IL-2). Cells at1 × 106/ml were cultured with irradiated (50 Gy) tumourcells at a ratio of 50 lymphocytes to one tumour cell. After5 days of incubation cells were harvested supernatantswere collected and tested for cytokines (IFNg and IL-4)using ELISA kits from PharMingen, following the manu-facturer’s instructions. A standard cytotoxic T cell (CTL)assay was carried out by 51Cr release from labelledtumour cells and various effector (CTL) to target (tumourcells) ratios. To investigate infiltration of tumours by leu-kocytes, tumours were excised and passed through100 mm filters (Falcon) to obtain a single cell suspensionand then stained with antibodies for CD45, CD4, CD8,and CD56 (from Pharmingen) followed by FITC-labelledsecondary antibody. Cells were analysed on a FACScan(Becton-Dickinson) with 10 000 events collected per sample.

ApoptosisTumour cells with or without the tk gene were culturedwith 5 mg/ml GCV and then detached as above. Anyfloating cells were pooled together with cells detachedfrom plates with trypsin/EDTA. Cells were assayed forapoptosis using an APO-BrDU kit (Pharmingen) follow-ing the manufacturer’s instructions. Briefly, the cells werefixed in 1% paraformaldehyde, washed, and then storedin 70% ethanol at −20°C until assayed. The cells werewashed in PBS, resuspended in 50 ml of DNA labellingsolution containing 10 ml reaction buffer, 0.75 ml TdTenzyme, 8.0 ml BrDUTP and 32.25 ml distilled water. Thecells were incubated for 60 min at 37°C. The cells werestained with anti-BrDU antibody and then counterstainedwith propidium iodide shortly before running on a FAC-Scan. 10 000 events were collected per sample.

Magnetic resonance spectroscopyAnimals were placed in the bore of a Varian 200–300spectrometer at 4.7T with the tumour resting on a 10 mmdual-turn 31P surface coil. 31P MR spectra were acquiredusing image-guided localised spectroscopy by the ISIS(Image Selected In vivo Spectroscopy) technique38 withadiabatic pulses, 1024 scans, a spectral width of 5000 Hzand a repetition time of 3 s. A voxel for localised 31Pspectroscopy was prescribed from a 1H scout image avo-iding animal body wall where this was evident in theimage. 10 Hz line-broadening was applied before Fouriertransform. The position of the coil was marked on eachtumour with indelible ink following the first experimentsession to ensure the same tumour position for sub-sequent experiments. Spectra were analysed using VAR-PRO, a variable projection time-domain non-linear leastsquares fitting algorithm,39 implemented using Matlabversion 5.1 (The Mathworks Inc., MA, USA). The relativepeak areas of visible 31P-containing metabolites weredetermined for each spectrum and various ratios arereported. Total phosphate (SP) for each spectrum wascalculated as the sum of the areas of all the 31P-contain-ing peaks in that spectrum.


567StatisticsMean tumour volumes were compared between groupsusing the Student’s t test. Survival between groups wasanalysed by the log-rank test.

AcknowledgementsJD Eaton was supported by the Swire Group. MJA Perrywas supported by the British Urological Foundation andOnyvax Ltd. RA Mazucco was supported by a grant fromthe Cancer Research Campaign UK (CRC) SP1971/0404.SM Todryk was supported by Onyvax Ltd. We are grate-ful to Dr R Vile (Mayo Clinic, Rochester, USA) for provid-ing the HSVtk retroviral packaging line, and to Dr HPandha and Dr M Stubbs (St George’s Hospital MedicalSchool) for very helpful discussion.

References1 Hrouda D, Perry M, Dalgleish AG. Gene therapy for prostate

cancer. Semin Oncol 1999; 26: 455–471.2 Pardoll DM. Cancer vaccines. Nat Med 1998; 4: 525–531.3 Simons JW et al. Induction of immunity to prostate cancer anti-

gens: results of a clinical trial of vaccination with irradiatedautologous prostate tumor cells engineered to secrete granulo-cyte–macrophage colony-stimulating factor using ex vivo genetransfer. Cancer Res 1999; 59: 5160–5168.

4 Tjoa BA et al. Follow-up evaluation of prostate cancer patientsinfused with autologous dendritic cells pulsed with PSMA pep-tides. Prostate 1997; 32: 272–278.

5 Vesalainen S, Lipponen P, Talja M, Syrjanen K. Histologicalgrade, perineural infiltration, tumour-infiltrating lymphocytesand apoptosis as determinants oflong-term prognosis in pros-tatic adenocarcinoma. Eur J Cancer 1994; 30A: 1797–1803.

6 Hrouda D et al. Immunotherapy of advanced prostate cancer: aphase I/II trial using Mycobacterium vaccae (SRL172). Br J Urol1998; 82: 568–573.

7 Maraveyas A et al. Possible improved survival of patients withstage IV AJCC melanoma receiving SRL 172 immunotherapy:correlation with induction of increased levels of intracellularinterleukin-2 in peripheral blood lymphocytes. Ann Oncol 1999;10: 817–824.

8 Heriot AG et al. Reduction in cytokine production in colorectalcancer patients: association with stage and reversal by resection.Br J Cancer 2000; 82: 1009–1012.

9 Blades RA et al. Loss of HLA class I expression in prostate can-cer: implications for immunotherapy. Urology 1995; 46: 681–687.

10 Freeman SM et al. The ‘bystander effect’: tumor regression whena fraction of the tumor mass is genetically modified. Cancer Res1993; 53: 5274–5283.

11 Vile RG et al. Systemic gene therapy of murine melanoma usingtissue specific expression of the HSVtk gene involves animmune component. Cancer Res 1994; 54: 6228–6234.

12 Ramesh R et al. In vivo analysis of the ‘bystander effect’: a cyto-kine cascade. Exp Hematol 1996; 24: 829–838.

13 Alvarez RD, Curiel DT. A phase I study of recombinant aden-ovirus vector-mediated intraperitoneal delivery of herpes sim-plex virus thymidine kinase (HSV-TK) gene and intravenousganciclovir for previously treated ovarian and extraovarian can-cer patients. Hum Gene Ther 1997; 8: 597–613.

14 Sterman DH et al. Adenovirus-mediated herpes simplex virusthymidine kinase/ganciclovir gene therapy in patients withlocalized malignancy: results of a phase I clinical trial in malig-nant mesothelioma. Hum Gene Ther 1998; 9: 1083–1092.

15 Herman JR et al. In situ gene therapy for adenocarcinoma ofthe prostate: a phase I clinical trial. Hum Gene Ther 1999; 10:1239–1249.

16 Pettaway CA et al. Selection of highly metastatic variants of dif-ferent human prostatic carcinomas using orthotopic implan-tation in nude mice. Clin Cancer Res 1996; 2: 1627–1636.

Gene Therapy

17 Stubbs M, Griffiths JR. Monitoring cancer by magnetic reson-ance. Br J Cancer 1999; 80 (Suppl. 1): 86–94.

18 Griffiths JR et al. 31P-NMR investigation of solid tumours in theliving rat. Biosci Reps 1981; 1: 319–325.

19 Griffiths JR et al. 31P-NMR studies of a human tumour in situ.Lancet 1983; i: 1435–1436.

20 deCertaines JD et al. In vivo 31P MRS of experimental tumours.NMR Biomed 1993; 6: 345–365.

21 Ross BD, Kim B, Davidson B. Assessment of ganciclovir toxicityto experimental intracranial gliomas following recombinant aden-oviral mediated transfer of the herpes simplex virus thymidinekinase gene by magnetic resonance imaging and proton magneticresonance spectroscopy. Clin Cancer Res 1995; 1: 651–657.

22 Izquierdo M et al. Long term rat survival after malignant brain-tumor regression by retroviral gene-therapy. Gene Therapy 1995;2: 66–69.

23 Namba H et al. Evaluation of the bystander effect in experi-mental brain tumors bearing herpes simplex virus-thymidinekinase gene by serial magnetic resonance imaging. Hum GeneTher 1996; 7: 1847–1852.

24 Stegman LD, Ben-Yoseph O, Freyer JP, Ross BD. In vivo 31P MRSevaluation of ganciclovir toxicity in C6 gliomas stably express-ing the herpes simplex thymidine kinase gene. NMR Biomed1996; 9: 364–368.

25 Hakumäki JM et al. 1H MRS detects polyunsaturated fatty acidaccumulation during gene therapy of glioma: implications forthe in vivo detection of apoptosis. Nat. Med. 1999; 5: 1323–1327.

26 Melcher A et al. Tumor immunogenicity is determined by themechanism of cell death via induction of heat shock proteinexpression. Nat. Med. 1998; 4: 581–587.

27 Hrouda D et al. Mycobacterium vaccae (SRL172): a potentialimmunological adjuvant evaluated in rat prostate cancer. Br JUrol 1998; 82: 870–876.

28 Hrouda D et al. Allogeneic whole tumour cell vaccination in therat model of prostate cancer. Br J Urd 2000; 86: 742–748.

29 Vieweg J et al. Immunotherapy of prostate cancer in the Dun-ning rat model: use of cytokine gene modified tumor vaccines.Cancer Res 1994; 54: 1760–1765.

30 Sanda MG et al. Demonstration of a rational strategy for humanprostate cancer gene therapy. J Urol 1994; 151: 622–628.

31 Eastham JA et al. Prostate cancer gene therapy: herpes simplexvirus thymidine kinase gene transduction followed by ganciclo-vir in mouse and human prostate cancer models. Hum Gene Ther1996; 7: 515–523.

32 Vile RG et al. Generation of an anti-tumour immune responsein a non-immunogenic tumour: HSVtk killing in vivo stimulatesa mononuclear cell infiltrate and a Th1-like profile of intratum-oural cytokine expression. Int J Cancer 1997; 71: 267–274.

33 Vile RG, Diaz RM, Castleden S, Chong H. Targeted gene therapyfor cancer: herpes simplex virus thymidine kinase gene-mediated cell killing leads to anti-tumour immunity that can beaugmented by co-expression of cytokines in the tumour cells.Biochem Soc Trans 1997; 25: 717–722.

34 Steen RG et al. 31P NMR spectroscopic and near infrared spec-trophotometric studies of effects of anesthetics on in vivo RIF-1tumors. Relationship to tumor radiosensitivity. NMR Biomed1989; 2: 87–92.

35 Rodriguez LM et al. 31P nuclear magnetic resonance spec-troscopy studies of the response of rat mammary tumours toendocrine therapy. Cancer Res 1988; 48: 89–93.

36 Negendank WG. Studies on human tumours by MRS: a review.NMR Biomed 1992; 5: 303–324.

37 Ruiz-Cabello J, Cohen JS. Phospholipid metabolites as indicatorsof cancer cell function. NMR Biomed 1992; 5: 226–233.

38 Ordidge RJ, Connelly A, Lohman JAB. Image-selected in vivospectroscopy (ISIS). A new technique for spatially selectiveNMR spectroscopy. J Magn Reson 1986; 66: 283–294.

39 van der Veen JWC, de Beer R, Luyten PR, Van Ormondt D.Accurate quantification of in vivo P-31 NMR signals using thevariable projection method and prior knowledge. Magn Res Med1988; 6: 92–98.

research article genetic prodrug activation therapy (gpat) in...

Documents