HSV vector cytotoxicity is inversely correlated with effective ... - Nature

Gene Therapy (2000) 7, 1483–1490  2000 Macmillan Publishers Ltd All rights reserved 0969-7128/00 $15.00 www.nature.com/gt

ACQUIRED DISEASES

RESEARCH ARTICLE

HSV vector cytotoxicity is inversely correlated with effective TK/GCV suicide gene therapy of rat gliosarcoma S Moriuchi1,2, DM Krisky1, PC Marconi1,3, M Tamura1, K Shimizu2, T Yoshimine2, JB Cohen1 and JC Glorioso1 1 2

Department of Molecular Genetics and Biochemistry, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA; and Department of Neurosurgery, Osaka University Medical School Osaka, Japan

Herpes simplex virus (HSV)-mediated delivery of the HSV thymidine kinase (tk) gene to tumor cells in combination with ganciclovir (GCV) administration may provide an effective suicide gene therapy for destruction of malignant glioblastomas. However, because HSV is a highly cytotoxic agent, gene expression from the virus is short-lived which may limit the effectiveness of HSVtk/GCV therapy. Using different replication-defective HSVtk gene vectors, we compared HSV vector backgrounds for their cytotoxic activity on infection of 9L gliosarcoma cells in culture and brain tumors in rats and evaluated the impact of vector toxicity on the effectiveness of tk/GCV-mediated suicide gene therapy. As reported previously for other cell lines, a vector deleted for both copies of the immediate–early (IE) gene ICP4 (SOZ.1) was highly toxic for 9L cells in culture while a vector deleted in addition for the ICP22 and ICP27 IE genes (T.1) reduced or arrested

9L cell proliferation with more limited cell killing. Nevertheless, both vectors supported widespread killing of uninfected cells in the presence of GCV following low multiplicity infections, indicating that vector cytotoxicity did not preempt the production of vector-encoded TK enzyme necessary for the killing of uninfected cells by the HSV-tk/GCV bystander effect. Although an SOZ.1-related vector (SHZ.2) caused tumor cell necrosis in vivo, injection of SHZ.2 at multiple coordinates thoughout the tumor followed by GCV administration failed to prolong markedly the survival of tumorbearing rats. In contrast, a single injection of T.1 produced a life-extending response to GCV. These results indicate that vector cytotoxicity can limit the efficacy of HSV-tk/GCV treatment in vivo, which may be due to premature termination of tk gene expression with attendant abortion of the bystander effect. Gene Therapy (2000) 7, 1483–1490.

Keywords: herpes simplex virus; gene therapy; glioblastoma; cytotoxicity; HSV thymidine kinase; ganciclovir

Introduction Despite recent advances in the treatment of malignant gliomas, the prognosis for patients suffering from this disease continues to be poor.1–3 Current therapies involve the use of surgery, radiotherapy, and chemotherapy, but these measures are inadequate due to the significant morbidity associated with these procedures, the infiltrative nature of gliomas, and the development of drug resistance.4–6 The challenges of malignant gliomas are unique given the immuno-privileged status of the central nervous system and the confined and sensitive nature of the tumor environment. A novel approach to the treatment of these tumors is the use of virus-based gene therapeutic vectors. The aim of these approaches is to transduce tumors with genes that can aid in the specific destruction of tumor cells. For example, an antitumor immune reac-

Correspondence: JC Glorioso, Department of Molecular Genetics and Biochemistry, E1240 Biomedical Science Tower, University of Pittsburgh School of Medicine, Pittsburgh, PA 15261, USA 3 Current address: Department of Experimental and Diagnostic Medicine (Section of Microbiology) and Biotechnology Center, University of Ferrara, Ferrara, Italy Received 15 March 2000; accepted 30 May 2000

tion may be triggered upon delivery of immunostimulatory gene products although immunosuppressive substances commonly secreted by gliomas7,8 may hinder this process. Another strategy involves blocking DNA synthesis through the delivery of ‘suicide genes’ such as those encoding herpes simplex virus thymidine kinase (HSV-TK), cytosine deaminase (CD), or cytochrome p450 2B1. The products of these genes have the ability to convert certain nontoxic pro-drugs to activated forms which interfere with DNA synthesis. Most cells in the central nervous system are nonreplicating and terminally differentiated suggesting that local gene therapy leading to the cessation of DNA synthesis should be largely tumor cell specific. Previous experiments using various viral vectors to deliver the HSVtk gene as part of a suicide gene therapy strategy have yielded generally positive results in a number of glioma models.9–12 The HSV-TK protein phosphorylates the nontoxic pro-drug ganciclovir (GCV), which results in the generation of a defective nucleoside analogue. This product, when incorporated into cellular DNA during replication, leads to DNA chain termination and selective killing of dividing cells. Of particular interest in this strategy is the ability of the defective nucleoside analogue to be passed to cells that are in physical

Effect of HSV vector toxicity in TK/GCV therapy S Moriuchi et al

1484

contact with the GCV-converting cells. This leads to widespread killing of both the transduced cells and neighboring cells, a phenomenon referred to as the ‘bystander effect’.13 The bystander effect allows the possibility of complete tumor eradication even at relatively low levels of target cell transduction. In this report, we examined the utility of replicationdefective mutant HSV-1 vectors for gene therapeutic approaches to malignant gliomas. Defective vectors have been produced which can safely transduce various regions of the brain with minimal toxicity for the host cell.14 These vectors are based upon an HSV-1 (KOS) mutant, d120, which is deleted for both copies of the essential immediate–early (IE) gene ICP4.15 The deletion of this gene, whose product is a major transactivator of HSV-1 transcription, results in a vector backbone that produces other proteins of the IE temporal class, but genes of the early and late classes are dramatically reduced in expression.16 Promoters of the IE type can be used to express transiently transgenes at high levels.17 We have derived a replication-deficient HSV-1 vector from d120 that uses the strong IE promoter of the ICP4 gene to drive expression of the HSVtk gene (SOZ.1) and have tested this vector as a therapeutic agent in a rat gliosarcoma model. We found that SOZ.1 was toxic for 9L gliosarcoma cells, yet supported enhanced tumor cell killing in the presence of GCV in vitro. In vivo experiments indicated, however, that viral cytotoxicity limited the efficacy of the HSV-TK/GCV suicide system. By deleting two additional IE genes, those encoding ICP22 and ICP27, vectors displaying substantially reduced cytotoxicity in vitro were created.18,19 These triple IE gene-deficient vectors showed superior performance in prolonging the survival of 9L tumor-bearing rats in a GCV-dependent manner.

Results Recombinant viruses Vectors SOZ.1, THZ.1, T.1 and TOZ.1 used in this work (Figure 1) have been described previously.18 Briefly, all

Figure 1 Schematic representation of replication-defective HSV-1-based HSV-TK vectors. Repeat regions are illustrated as open boxes and IE genes inactivated by deletion as black bars. Hatched boxes represent LacZ expression cassettes and shaded boxes the ICP4 promoter (ICP4p) introduced in front of the native HSVtk gene. The positions of pertinent HSV loci are indicated at the bottom. Gene Therapy

four had both copies of the ICP4 gene deleted and the native early (E) promoter of the HSVtk gene (UL23) substituted by the ICP4 immediate–early (IE) promoter of HSV-1 to assure synthesis of the HSV-TK protein under conditions where the normal progression from IE to E to late (L) gene expression is blocked by the absence of the ICP4 protein, an essential transactivator of HSV E and L gene transcription. This promoter substitution also inactivated the UL24 gene whose coding region partially overlaps with the tk promoter. Three of the four recombinants contained the lacZ reporter gene encoding ␤-galactosidase controlled by either the strong human cytomegalovirus (HCMV) IE promoter (THZ.1) or the HSV-1 ICP0 IE promoter (SOZ.1 and TOZ.1). Three of the viruses (THZ.1, T.1 and TOZ.1) had additional deletions inactivating the IE genes ICP27, an essential gene, and ICP22. The fifth vector, SHZ.2, has also been previously described.18 SHZ.2 contained an HCMV IEp-lacZ expression cassette and an ICP4 IEp-tk cassette as a Crelox insertion in UL23 of an ICP4-deleted virus. In the nomenclature of these viruses, S stands for single (ICP4) and T for triple (ICP4, 22, 27) IE gene deletion. It was previously reported that combined IE gene deficiencies substantially reduced the cytotoxicity of HSV-1 and ICP4deficient derivatives,18–22 which allowed for extended transgene expression.18,19,22

Vector cytotoxicity Cell viability assays were performed to assess the toxicity of viral mutants deleted for one or three IE genes. Rat gliosarcoma 9L cells which are highly susceptible to infection with replication-incompetent ICP4− HSV-1 mutants18 were infected with the single or triple IE gene deletion viruses SOZ.1, T.1 and TOZ.1 at multiplicities of infection (MOIs) of 0.1, 1 and 10. Growth rates were measured by MTT assay, transduction rates and morphologies of SOZ.1- and TOZ.1-infected cells analyzed by Xgal staining for ␤-galactosidase activity, and apoptosis examined by a fluorescent TUNEL assay. Infection of 9L cells with SOZ.1 at an MOI of 10 resulted in cell loss by 3 days post-infection (p.i.) (Figure 2), at which time a majority of the cells (98%, Figure 3) were undergoing apoptosis. In contrast, while infection

Figure 2 Survival and growth of SOZ.1- and T.1-infected 9L gliosarcoma cells in culture. The cells were infected at MOIs of 0.1, 1 or 10 and relative cell densities were determined by MTT assay.


1485

Figure 3 Comparison of apoptosis induced by SOZ.1, T.1 and TOZ.1 in cultured 9L cells. Cells were infected at multiplicities of 1, 5 or 10, and the percentage of apoptotic cells 3 days p.i. determined by a fluorescent TUNEL assay.

with T.1 at the same MOI caused a reduction in cell growth (Figure 2), most of the cells were viable at 3 days p.i. with only 25% undergoing apoptosis (Figure 3). Likewise, TOZ.1-infected cultures displayed just 25% apoptotic cells. Minimal estimates of transduction efficiencies were obtained by counting X-gal-positive and -negative cells following infection with SOZ.1 and TOZ.1 and were found to be reflective of virus dose. Thus, all cells were ␤-galactosidase positive at high MOI, whether infected with SOZ.1 or TOZ.1, while infection at multiplicities of 1 and 0.1 showed blue staining of approximately 70% and 10% of the cells, respectively, consistent with a Poisson distribution (Figure 4a and data not shown). At 3 days p.i., a majority of the cells in cultures infected with SOZ.1 at high MOI were rounded or irregularly shaped, indicative of viral toxicity, while most cells in T.1- or TOZ.1infected cultures were polygonal in shape and appeared healthy (Figure 4b). Likewise, many SOZ.1-infected cells in the X-gal stained cultures (1 day p.i.) were rounded in shape compared with the predominantly polygonal shape of TOZ.1-infected cells (Figure 4a). Inhibition of cell growth by SOZ.1 and T.1 infections was also observed at lower MOIs (Figure 2), but the differences were less apparent than at the highest MOI. At the same time, a clear difference remained in the percentage of apoptotic cells in SOZ.1-infected cultures (50%) compared with T.1- (20%) and TOZ.1-infected cultures (10%) at an MOI of 1 3 days p.i. These results demonstrated that vectors deleted for ICP4 alone are cytotoxic for 9L cells in vitro while triple IE gene deletion viruses are less toxic, although equally infectious.

Efficacy of GCV treatment To explore the effect of virus cytotoxicity on cell killing in the presence of ganciclovir (GCV), 9L cells were infected at different MOIs with SOZ.1 or T.1, and exposed to GCV for 3 days. While SOZ.1 infection again reduced the number of viable cells more than T.1 infection at the higher MOIs without GCV, no clear differences were observed at any MOI in the presence of GCV

Figure 4 (a) SOZ.1- and TOZ.1-infected 9L cultures (MOI = 0.1 or 10) stained at 24 h p.i. for ␤-galactosidase activity. (b) Phase-contrast photographs of 9L cells infected with single or triple IE gene deletion HSV vectors (MOI = 10; 3 days p.i.).

(Figure 5). Thus, no viable cells remained in the GCVtreated cultures infected at the higher MOIs, and similar fractions of the cells remained at the lowest MOI. In the absence of GCV, approximately 90% of the cells infected at the lowest MOI with either virus survived, as before (Figure 2), but a strong bystander effect of GCV treatment was apparent as far larger fractions of these surviving cells were killed by exposure to GCV (approximately 90%) than would be expected if TK-activated GCV killed only infected cells (⬍10% at an MOI of 0.1). These results indicated that GCV treatment of HSVtk-transduced cells can be effective in vitro whether or not the vector is cytotoxic.

Therapeutic value of virus cytotoxicity Animal survival experiments were performed to determine whether SOZ.1 alone or in combination with GCV treatment could impair tumor growth. Rat intracranial implants of 9L cells were injected after 5 days with 1 × 107 p.f.u. of SOZ.1, and GCV was administered daily for Gene Therapy


1486

7 days to half the animals by intraperitoneal injection. As illustrated in Figure 6, SOZ.1 failed to prolong the survival of tumor-bearing rats compared with untreated controls, and GCV administration did not alter the outcome. The lack of GCV effect was consistent with the interpretation that the intended source of HSV-TK, the fraction of tumor cells transduced by the vector, had been rapidly eliminated by the cytotoxicity of the vector, thus precluding the development of a bystander effect. At the same time, the infection rate was apparently insufficient to achieve a significant delay in tumor growth by viral cytotoxicity alone. To examine virus cytotoxicity in vivo, the SOZ.1-related vector SHZ.2 (see Figure 1) was injected into intracranial 9L tumors and contralateral control sites 7 days after tumor cell implantation (3 × 107 p.f.u.). Histopathologic analysis of sections taken 2 days later showed tumor cell necrosis in the immediate area of injection (Figure 7b;

compared with uninjected tumor, Figure 7a) while no damage was apparent in adjacent brain tissue. No areas of necrosis were observed in contralateral sections (data not shown). These results showed that the virus was toxic for intracranial 9L tumor cells, but also indicated that the

a

b

Figure 5 Survival of infected 9L cells in the presence of GCV. Cultured 9L cells were infected with SOZ.1 or T.1 at the indicated MOIs and incubated with or without GCV (10 ␮g/ml). At 3 days p.i., cell densities were determined by MTT assay. The results are shown as survival relative to untreated, mock-infected cells. The bars represent the averages of triplicate determinations. Standard deviations that were large enough to be distinguishable in the figure are illustrated.

Figure 6 Kaplan–Meier survival plot of Fisher 344 rats with intracerebral 9L tumors treated by a single injection of SOZ.1 virus with or without daily intraperitoneal GCV administration (30 mg/kg) for 7 days. The key shows the median survival times in days and the number of animals per treatment (n). Gene Therapy

c

Figure 7 In vivo cytotoxicity of SHZ.2. Fisher 344 male rats were inoculated with 4 × 104 9L cells in the frontal cortex 7 days before intratumoral virus injection. Using stereotactic apparatus, 3 × 107 p.f.u. of virus was delivered by a single 15 ␮l injection at the tumor coordinates or 15 1 ␮l injections at and around the same coordinates within a 2 × 2 × 2 mm area representing the confines of tumors 7 days after implantation. Two days after virus injection, the rats were perfused with PBS followed by 25% paraformaldehyde in PBS, and brains were removed, sectioned, and stained with hematoxylin and eosin. (a) Uninjected tumor; (b) single virus injection; (c) multipoint virus injection.


area of infection was too limited to control tumor growth in the absence of a bystander effect. The procedure, using a single injection of replication-incompetent virus, left normal surrounding cells unaffected. The lack of GCV effect on the survival of tumor-bearing, SOZ.1-injected animals (Figure 6) and the demonstrated tumor cell necrosis at the sites of SHZ.2 injection (Figure 7b) supported the suggestion that virus cytotoxicity interferes with the production of HSV-TK and thus with the development of a bystander effect. As a potential alternative to the bystander effect, we examined whether high-density delivery of one of these cytotoxic vectors (SHZ.2) could be an effective antitumor weapon. Intracranial 9L tumors approximately 8 mm3 in size (7 days after implantation) were injected at 15 coordinates with the same total amount of virus as applied earlier by single injection (3 × 107 p.f.u.), and sections were examined 2 days later. As shown in Figure 7c, multipoint injection caused necrosis of the entire tumor area without damage to the adjacent normal brain tissue. Thus, the virus appeared to be capable of eliminating the bulk of the tumor when administered at high density. Nevertheless, multipoint injection of SHZ.2 with or without GCV treatment failed to extend the life-span of tumor-bearing rats by more than a few days at best (Table 1, 7 days after implantation), even when performed at an early stage of tumor development (3 days after implantation; Table 1). From these results we concluded that the cytotoxicity of single IE-gene deletion viruses (SOZ.1, SHZ.2) had no practical therapeutic value and that these vectors were ineffective agents of suicide gene therapy.

Therapeutic value of a less cytotoxic virus The preceding results suggested that less cytotoxic vectors such as T.1 would be more effective tools for HSVTK/GCV therapy than such highly cytotoxic vectors as SOZ.1. To test this suggestion, intracranial 9L tumors were injected with T.1 virus 5 days after implantation and exposed to GCV treatment; virus delivery was by a single injection of 1 × 107 p.f.u. and GCV was administered intraperitoneally on 10 successive days. Figure 8 shows that while the individual and median survival times of medium- and T.1-injected animals were similar, GCV administration now provided a clear benefit (P ⬍ 0.05), delaying the demise of T.1-injected animals by up

Table 1 Survival of rats injected with SHZ.2 on days 3 or 7 following 9L tumor cell implantationa Vector injectionb

Treatment

Mean survival (days)

Day 3

Medium Medium + GCVc SHZ.2 SHZ.2 + GCVc

19.2 20.8 22.4 23.6

Day 7

Medium Medium + GCVc SHZ.2 SHZ.2 + GCVc

17.6 16.6 21.6 18.6

a

SHZ.2 was administered to the tumor area by multiple injections (3 × 107 p.f.u. total). b Days after tumor cell implantation. c GCV was administered on 4 successive days starting with the day of vector delivery.

1487

Figure 8 Kaplan–Meier survival plot of Fisher 344 rats with intracerebral 9L tumors treated by a single injection of T.1 virus with or without daily GCV administration for 10 days. The key shows the median survival times in days and the number of animals per treatment.

to 15 days or more for the longest survivors. This result confirmed that vector cytotoxicity limited the effectiveness of HSV-based antitumor suicide gene therapy, presumably by shutting down the sites of HSV-TK production, and demonstrated that this limitation could be resolved by deleting a combination of immediate–early genes from the vector. In all, this study showed that bystander killing is a more effective mechanism against intracranial 9L tumors than vector cytotoxicity, at least when the vector is replication incompetent to prevent uncontrolled virus spread.

Discussion The hit-and-run mechanism by which cytocidal viruses amplify while destroying the host cell might be an ideal antitumor weapon if the infectious specificity of the virus could be restricted to the tumor cell. Many viruses require active cell division for replication, thus favoring tumor cells over terminally differentiated normal cells as targets, but virus release into the circulation is not easily contained and can cause serious side-effects. Although replication-competent vectors are theoretically more efficient in infecting the majority of tumor cells in a solid tumor than replication-defective vectors, this remains to be formally demonstrated. Replication-defective vectors offer a better safety profile because they are not amplified in vivo and do not readily spread beyond the initial site of infection, especially when injected into a region with limited access to the circulation. However, this inability to spread also allows tumor cells to escape infection, thus limiting the effectiveness of replication-defective cytotoxic vectors. This limitation may be obviated by the use of vectors that supply gene products such as HSV-TK that are capable of expanding the field of tumor cell killing to noninfected cells. We have compared two mechanisms of virus-dependent tumor destruction using replication-incompetent HSV vectors and have shown that only one is effective, but not in combination with the other. To evaluate the Gene Therapy


1488

first mechanism, direct virus-mediated cell killing, cytotoxic virus was distributed throughout small tumors by injection at numerous sites. While this procedure resulted in widespread tumor cell necrosis, lethal tumor outgrowth was not substantially delayed indicating that a portion of the tumor cells had escaped infection. The second mechanism did not require complete infection of the tumor mass, but instead relied on the spread of viral TK-activated GCV to reach uninfected tumor cells. However, when GCV was administered after intra-tumoral delivery of a pro-drug activating but cytotoxic HSVtk gene vector, no significant increase in animal survival was observed. On the other hand, GCV had a markedly beneficial effect on animals injected with a less cytotoxic TK vector, indicating that vector cytotoxicity could limit the production of TK enzyme and hence the production and spread of toxic GCV. We concluded that vector cytotoxicity interfered with effective HSV-TK/GCV therapy using replication-defective TK vectors and that the deletion of multiple viral IE genes reduced this deficiency. It has been repeatedly demonstrated in the past that HSV vector cytotoxicity precludes prolonged transgene expression.18,19,22 Several of the immediate–early genes of HSV are cytotoxic when expressed separately20 and their removal from the virus genome allows protracted synthesis of viral and transgene products provided that the promoters controlling these genes function independently of IE proteins, in particular of the transcriptional transactivator ICP4. The IE promoters are convenient examples of IE protein-independent promoters17 and we used the ICP4 promoter in our vectors to direct HSVtk gene expression. In agreement with previous studies of HSV vector toxicity and transgene expression in different cell types,18,19,22–24 we observed that removal of the ICP22 and ICP27 genes from an ICP4-deficient vector background improved the response of vector-injected 9L gliosarcomas to GCV. Aside from the cytotoxic IE genes, HSV-1 infection also introduces into the host cell the UL41 vhs gene product, a component of the viral tegument causing host proteinsynthesis shut-off which may contribute to the cytotoxicity of replication-deficient HSV vectors.25 Our vectors T.1 and TOZ.1 differed exclusively in the UL41 locus, which was wild-type in T.1, but disrupted by a lacZ expression cassette in TOZ.1. These two vectors had similar effects in our apoptosis assay, suggesting that the vhs gene product contributes little to the cytotoxicity of replicationdefective HSV vectors, in agreement with previous observations.18,20,21 Accordingly, inactivation of the UL41 locus is not a critical step in HSV vector design although the locus is a convenient site for the insertion of therapeutic transgenes.26 Our in vitro studies showed that infection of 9L cells with the single IE gene deletion vector SOZ.1 at high MOI, leading to infection of a majority of the cells, caused a net loss of viable cells as early as 2 days p.i., pointing not merely to inhibition of cell proliferation, but to rapid cell killing. Indeed, a large fraction of these cells were undergoing apoptosis 3 days after infection and infected cells displayed morphological changes indicative of severe stress. Nevertheless, a sizable bystander effect was observed in SOZ.1-infected cultures, suggesting that HSV-TK was synthesized early enough relative to virusmediated cell death to allow adequate production and

Gene Therapy

distribution of toxic GCV. Thus, although vector cytotoxicity appeared to preclude a significant HSV-TK/GCV antitumor effect, it was compatible with HSV-TK/GCV bystander killing in vitro. This discrepancy may relate to the effective availability of GCV, which is instantly adequate in vitro, but not when the pro-drug is administered by i.p. injection in vivo. Unlike the experimental animals used in this work, a large fraction of the human population has been previously exposed to HSV infection, which raises the likelihood of local inflammation around the tumor site in response to virus injection. With other vector systems delivering HSV-TK to 9L tumors, correlations have been noted between nonspecific local inflammation and the development of a specific antitumor immune response.27 This suggests the possibility that the local therapeutic effects of HSV-based TK vectors in combination with GCV may be enhanced by pre-exposure to HSV. While an anti-HSV response is generally undesirable in HSVbased gene therapy applications, the possibility that it positively contributes to antitumor therapy deserves further experimental scrutiny. On the other hand, a recent study reported widespread chronic brain inflammation, vector distribution, and detection of TK protein in animals cured of experimental gliomas by local injection of replication-incompetent adenoviral TK vectors in combination with systemic GCV treatment.28 Hence, the occurrence of acute and chronic inflammation and their effects in response to HSV-based TK/GCV therapy are important areas of future investigation. In summary, our results show that replication-defective HSV vectors become better tools for HSV-TK/GCV therapy when their cytotoxicity is reduced, which can be accomplished by a combination of IE gene deletions. While vector cytotoxicity did not abolish the HSV-TK bystander effect in vitro, it did in vivo, presumably because the transduced effector cells were incapacitated before sufficient amounts of toxic GCV metabolite could be produced and exported. Future directions to improve the success of tumor treatment using HSV-based gene therapy vectors include optimizing the relative timing of vector and GCV administration, exploring the effects of previous exposure to HSV, and the development of vectors that use tumor cell-specific receptors for infection and express transgenes that improve bystander killing. For example, we have recently demonstrated in animal models that radiosurgery combined with vector-directed production of tumor necrosis factor alpha can improve the efficacy of HSV-TK/GCV therapy for glioblastoma34 and that expression of connexin 43 from an HSV-TK vector improves the GCV bystander effect in tumors with reduced intercellular communication through gap junctions.29 These findings, together with our present results, suggest that multi-modal therapies combining conventional treatments with the use of highly defective vectors capable of delivering multiple therapeutic gene products may provide unparalleled new opportunities for effective clinical treatment of recurrent glioma.

Materials and methods Cell culture and virus production Vero African green monkey kidney cells (CCL81; ATCC, Rockville, MD, USA) and Vero-derived E515 and 7B cells30


were maintained in MEM (LTI, Gaithersburg, MD, USA) supplemented with glutamine, antibiotics (PSN), and 10% fetal bovine serum (all from LTI). 9L rat gliosarcoma cells (kindly provided by Dr KM Culver, NIH, Bethesda, MD, USA) were maintained in DMEM supplemented with glutamine, antibiotics, and 10% fetal bovine serum. HSV-1 stocks were prepared by infecting E5 or 7B cells, which supply the ICP4 protein, respectively, the ICP4 and ICP27 proteins in trans,15,30 at a multiplicity of 0.03. Infected cells were harvested when a 100% cytopathic effect was evident and subjected to three cycles of freeze– thawing (−80°C/37°C) and a single burst of sonication. The virus was then separated from cell debris by centrifugation. Virus stocks were aliquoted into 2 ml cryogenic tubes (Corning Glass Works, Corning, NY, USA) and stored at −80°C. Final titers averaged 2 × 108 to 2 × 109 p.f.u./ml.

Construction of recombinant HSV-1 vectors SHZ.218 was created by Cre-lox recombination31 of a human cytomegalovirus (HCMV) IE promoter (IEp)-lacZ expression cassette together with an HSV-1 ICP4 promoter (ICP4p)-HSVtk cassette into the UL23 (HSVtk) locus of a virus backbone based on d120, a KOS strain mutant deleted for both copies of the ICP4 gene15 (Figure 1). SOZ.1 was created via homologous recombination replacing a UL41 SmaI fragment of S4TK, a UL24−:ICP4p-tk derivative of d120,26 with an ICP0p-lacZ expression cassette.18 THZ.1 had the S4TK backbone with the HCMV IEp-lacZ cassette replacing a portion of the ICP22 gene and the ICP27 gene inactivated by a deletion.18 T.1 was derived from THZ.1 by removal of the lacZ gene, leaving behind within the ICP22 gene the upstream and downstream regulatory sequences of the lacZ cassette.18 TOZ.1 was produced as a cross between SOZ.1 and T.1.18 All marker transfer reactions were carried out by standard calcium phosphate transfection of 5 ␮g viral DNA and 1 ␮g linear recombination plasmid.30 Viral crosses were performed by co-infection of monolayers of 7B cells30 at an MOI of 5 for each virus followed by harvesting at 18 h p.i. All resultant viral mutants were purified by three rounds of limiting dilution and their genome structures were verified by Southern blotting. In vitro cytotoxicity assays In vitro cytotoxicity was examined by standard MTT assay.32 Cells were infected in suspension for 1 h at different MOIs and dispensed into 96-well, flat-bottomed microtiter plates (Becton Dickinson, Franklin Lakes, NJ, USA) at a concentration of 1 × 104 cells in 100 ␮l per well. The cells were incubated with or without GCV (10 ␮g/ml) at 37°C in 5% CO2. On each day after infection (Figure 2) or 3 days p.i. (Figure 5), 25 ␮l 3-(4,5-dimethylthiazol-2-yl)-diphenyltetrazolium bromide (MTT) solution (5 mg/ml; Sigma Chemicals, St Louis, MO, USA) was added to each well and incubation at 37°C in 5% CO2 was continued for 30 min. Supernatants were removed and 100 ␮l dimethylsulfoxide was added to each well. After another 30 min incubation, absorbances were measured using an EL312e plate reader equipped with a 570 nm filter (Biotek Instruments, Winooski, VT, USA). Apoptosis was visualized by a fluorescent TUNEL assay using the manufacturer’s instructions (Boehringer Mannheim, Indianapolis, IN, USA).

X-gal staining of lacZ-transduced cells 9L cells were washed, trypsinized, counted, and resuspended at 5 × 105 cells/ml in 1 × MEM containing 10% fetal bovine serum. Cells (1 ml) were added to 500 ␮l medium containing the appropriate amount of recombinant virus (SOZ.1 or TOZ.1), agitated at 37°C for 1 h, and plated in six-well plates (Falcon, Franklin Lakes, NJ, USA). After 24 h, the monolayers were washed, fixed with 4% paraformaldehyde in PBS (pH 7.4) for 15 min, and stained with X-gal as previously described.33

1489

In vivo tumor implantation and treatment SOZ.1 vector (Figure 6): After anesthetization with ketamine (87 mg/kg) and rompun (13 mg/kg), 1 × 105 9L cells in a volume of 5 ␮l were injected stereotactically into the frontal lobe of 180–220 g male Fisher 344 rats at a location 2.2 mm anterior and 2.0 mm lateral of the bregma (n = 16). After 5 days, the animals were reinjected at the same coordinates with 1 × 107 p.f.u. of SOZ.1 (n = 10) or 10 ␮l MEM. GCV was administered daily (30 mg/kg) to five of the virus-injected animals intraperitoneally for 7 days. Animals were killed at the onset of neurological symptoms or other signs suggesting an inability to care for themselves, in compliance with the guidelines of the University of Pittsburgh Animal Care and Use Committee. T.1 vector (Figure 8): T.1 was administered as described for SOZ.1. The experiment included 19 animals (12 injected with virus, eight of these receiving GCV), and daily GCV injections were performed for 10 days. SHZ.2 vector (Figure 7): 4 × 104 9L cells were implanted intracranially in a volume of 5 ␮l. After 7 days, 3 × 107 p.f.u. SHZ.2 in 30 ␮l MEM were injected at the same site. Brains were removed 2 days later and sections prepared for histopathology.

Acknowledgements We wish to thank Kate Sullivan of the Department of Molecular Genetics and Biochemistry for technical assistance. This work was supported by NIH Grants GM34534 and AR44526 to JCG.

References 1 Arcicasa M et al. Results of three consecutive combined treatments for malignant gliomas. Ten-year experience at a single institution. Am J Clin Oncol 1994; 17: 437–443. 2 Mahaley MS Jr et al. National survey of patterns of care for brain-tumor patients. J Neurosurg 1989; 71: 826–836. 3 Ushio Y. Treatment of gliomas in adults. Curr Opin Oncol 1991; 3: 467–475. 4 Chang SM, Prados MD. Chemotherapy for gliomas. Curr Opin Oncol 1995; 7: 207–213. 5 Kaye AH, Laidlaw JD. Chemotherapy for gliomas. Curr Opin Neurol Neurosurg 1992; 5: 526–533. 6 Salcman M. Epidemiology and factors affecting survival. In: Appuzzo MLJ (ed). Malignant Cerebral Glioma. American Association of Neurological Surgeons, AANS Publications Committee: Park Ridge, IL, 1990, pp 95–110. 7 Hishii M et al. Human glioma-derived interleukin-10 inhibits antitumor immune responses in vitro. Neurosurgery 1995; 37: 1160–1167. Gene Therapy


1490

8 Ruffini PA et al. Factors, including transforming growth factor beta, released in the glioblastoma residual cavity, impair activity of adherent lymphokine-activated killer cells. Cancer Immunol Immunother 1993; 36: 409–416. 9 Barba D et al. Development of antitumor immunity for following thymidine kinase-mediated killing of experimental brain tumors. Proc Natl Acad Sci USA 1994; 91: 4348–4352. 10 Chen SH et al. Gene therapy for brain tumors: regression of experimental gliomas by adenovirus-mediated gene transfer in vivo. Proc Natl Acad Sci USA 1994; 91: 3054–3057. 11 Perez-Cruet M et al. Adenovirus-mediated gene therapy of experimental gliomas. J Neurosci Res 1994; 39: 506–511. 12 Ram Z et al. Therapy of malignant brain tumors by intratumoral implantation of retroviral vector-producing cells. Nature Med 1997; 12: 1354–1361. 13 Freeman SM et al. The ‘bystander effect’: tumor regression when a fraction of the tumor mass is genetically modified. Cancer Res 1993; 53: 5274–5283. 14 Fink DJ et al. In vivo expression of ␤-galactosidase in hippocampal neurons by HSV-mediated gene transfer. Hum Gene Ther 1992; 3: 11–19. 15 DeLuca NA, McCarthy AM, Schaffer PA. Isolation and characterization of deletion mutants of herpes simplex virus type 1 in the gene encoding immediate–early regulatory protein ICP4. J Virol 1985; 56: 558–570. 16 DeLuca NA, Schaffer PA. Activation of immediate–early, early, and late promoters by temperature-sensitive and wild-type forms of herpes simplex virus type 1 protein ICP4. Mol Cell Biol 1985; 5: 1997–2008. 17 Krisky DM et al. Development of herpes simplex virus replication defective multigene vectors for combination gene therapy applications. Gene Therapy 1998; 5: 1517–1530. 18 Krisky DM et al. Deletion of multiple immediate–early genes from herpes simplex virus reduces cytotoxicity and permits long-term gene expression in neurons. Gene Therapy 1998; 5: 1593–1603. 19 Wu N et al. Prolonged gene expression and cell survival after infection by a herpes simplex virus mutant defective in the immediate–early genes encoding ICP4, ICP27, and ICP22. J Virol 1996; 70: 6358–6368. 20 Johnson P, Wang M, Friedmann T. Improved cell survival by the reduction of immediate–early gene expression in replicationdefective mutants of herpes simplex virus type 1 but not by mutation of the viron host shutoff function. J Virol 1994; 68: 6347–6362.

Gene Therapy

21 Marconi P et al. Replication-defective HSV vectors for gene transfer in vivo. Proc Natl Acad Sci USA 1996; 93: 11319–11320. 22 Samaniego LA. Neiderhiser L, DeLuca NA. Persistence and expression of the herpes simplex virus genome in the absence of immediate–early proteins. J Virol 1998; 72: 3307–3320. 23 Akkaraju GR et al. Herpes simplex virus vector-mediated dystrophin gene transfer and expression in MDX mouse skeletal muscle. J Gene Med 1999; 1: 280–289. 24 Huard J et al. Gene transfer to muscle using herpes simplex virus-based vectors. Neuromusc Dis 1997; 7: 299–313. 25 Read GS, Frenkel N. Herpes simplex virus mutants defective in the virion-associated shutoff of host polypeptide synthesis and exhibiting abnormal synthesis of ␣ (immediate early) viral polypeptides. J Virol 1983; 46: 498–512. 26 Krisky DM et al. Rapid method for construction of recombinant HSV gene transfer vectors. Gene Therapy 1997; 4: 1120–1125. 27 Culver K et al. In vivo gene transfer with retroviral vector-producer cells for treatment of experimental brain tumors. Science 1992; 256: 1550–1552. 28 Dewey RA et al. Chronic brain inflammation and persistent herpes simplex virus 1 thymidine kinase expression in survivors of syngeneic glioma treated by adenovirus-mediated gene therapy: implications for clinical trials. Nature Med 1999; 5: 1256–1263. 29 Marconi P et al. Connexin43-enhanced suicide gene therapy using herpesviral vectors. Mol Ther 2000; 1: 71–81. 30 Krisky D et al. Development of replication-defective herpes simplex virus vectors. In: Robbins P (ed). Methods in Molecular Medicine, Gene Therapy Protocols. Humana Press: New Jersey, 1996, pp 79–102. 31 Gage PJ et al. A cell free recombination system for site-specific integration of multigenic shuttle plasmids into the herpes simplex virus type 1 genome. J Virol 1992; 66: 5509–5515. 32 Mosmann T. Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J Immunol Meth 1983; 65: 55–63. 33 Sanes JR, Rubenstein JL, Nicolas JF. Use of a recombinant retrovirus to study post-implantation cell lineage in mouse embryos. EMBO J 1986; 5: 3133–3142.

Reference added in proof 34 Niranjan A et al. Effective treatment of experimental glioblastoma by HSV vector-mediated TNFa and HSV-tk gene transfer in combination with radiosurgery and ganciclovir administration. Mol Ther (in press).