Local and systemic antitumor response after combined ... - Nature

Gene Therapy (1997) 4, 1246–1255  1997 Stockton Press All rights reserved 0969-7128/97 $12.00

Local and systemic antitumor response after combined therapy of mouse metastatic tumors with tumor cells expressing IFN-a and HSVtk: perspectives for the generation of cancer vaccines L Santodonato1, G D’Agostino1, SM Santini1, D Carlei1, P Musiani2, A Modesti2, P Signorelli2, F Belardelli1 and M Ferrantini 1 1

Laboratory of Virology, Istituto Superiore di Sanita`, Rome; and 2 Institute of Human Pathology, G D’Annunzio University, Chieti, Italy

In this study, we have evaluated the local versus systemic antitumor response in tumor-bearing mice subjected to a combined therapeutic regimen based on the injection of genetically modified Friend erythroleukemia cells (FLC) producing IFN-a and expressing the HSVtk (tk) gene, and we have investigated the host immune mechanisms involved in tumor rejection and development of antitumor immunity. Repeated subcutaneous (s.c.) injections of IFNtk-expressing tumor cells, followed by GCV administration, were effective in counteracting the growth of both contralateral parental tumors as well as visceral metastases, whereas similar treatments with control tk cells (ie nonproducing IFN) were ineffective. Morphologic analyses of the homolateral and contralateral tumor tissues and in vivo immunosuppression experiments with specific monoclonal antibodies revealed that both CD4+ and CD8+ T lymphocytes played essential roles in the generation of a

definite antitumor response after the combined therapeutic regimen. We have also compared the effectiveness of irradiated versus viable tumor vaccines coexpressing the two genes in the FLC model and in the poorly immunogenic, metastasizing TS/A adenocarcinoma tumor system. Repeated injections of high doses of irradiated IFN-a-tkexpressing tumor cells followed by GCV administration resulted in the cure of the majority of mice bearing established metastatic tumors, while repeated inoculations of the same number of viable tumor vaccines were much less effective. We conclude that: (1) IFN-a is an essential cofactor in the generation of a systemic antitumor immunity following the prodrug-induced tumor cell killing; (2) vaccines co-expressing an autotoxic gene and a cytokine gene may represent promising new tools for the treatment of some cancer patients.

Keywords: IFN-a; cancer gene therapy; HSVtk

Introduction Since the first report on the in vivo behavior of genetically modified mouse tumor cells producing IL-4 in 1989,1 an increasing number of studies with transduced cells producing different cytokines have been published2 and much attention has been given to the possible clinical applications for cytokine gene therapy of cancer. As a result of some general enthusiasm in this field, many clinical trials have been started using genetically modified tumor vaccines expressing certain cytokines. Although these experimental and clinical studies have provided important information on the spectrum of immunological responses potentially induced by cytokine-producing vaccines, it is generally assumed that further strategies based on combined therapeutic approaches should be designed in order to achieve a

Correspondence: M Ferrantini, Laboratory of Virology, Istituto Superiore di Sanita`, Viale Regina Elena 299, 00161 Rome, Italy Received 11 April 1997; accepted 20 June 1997

more pronounced response in the therapy of certain human cancers. Our laboratory has been especially interested in studying the mechanisms of the antitumor effect of type I interferon (IFN) in mice and more recently the in vivo behavior of various mouse metastatic tumor cell types producing murine IFN-a1.3–6 During a set of studies aimed at defining the optimal conditions for achieving complete tumor rejection by using IFN-producing tumor cells, we found that the best approach was based on the use of tumor vaccines co-expressing IFN-a and an autotoxic gene.7 Thus, we have recently reported that peritumoral treatment with IFN-a-producing tumor cells expressing the HSVtk gene followed by repeated GCV treatments resulted in the long-term tumor eradication and cure of immunocompetent mice bearing established metastatic Friend leukemia cells (FLC) tumors.7 The success of this combined therapy was based on its ability to induce a potent and long-lasting antitumor immune response, due to a strong synergistic effect between the IFN-a locally secreted by the IFN-producing-tk-expressing tumor cells and the subsequent GCV-induced killing more than to a

Antitumor therapy with tumor cells expressing IFN-a and HSVtk L Santodonato et al

classic ‘bystander effect’.8 This was indicated by the failure of the combined therapy to eradicate FLC tumors established in immunosuppressed nude mice and by the ability of cured immunocompetent animals to resist parental tumor challenge even several months after complete eradication of the original tumor.7 In the present study, we have evaluated the local versus systemic antitumor response in tumor-bearing mice subjected to this combined therapeutic regimen and we have investigated the host immune mechanisms involved in tumor rejection and development of antitumor immunity. We herein report that repeated s.c. injections of IFNtkexpressing tumor cells, followed by GCV administration, are very effective in counteracting the growth of both contralateral parental tumors as well as visceral metastases, and that both CD4+ and CD8+ T lymphocytes play an essential role in the generation of a definite antitumor response after the combined therapeutic regimen. Moreover, the results of some experiments using multiple vaccine doses indicate that under certain conditions, irradiated cells co-expressing the two genes are even more effective than their viable counterparts in suppressing tumor growth. These results may provide an important insight into the definition of new strategies for the preparation of safe and more effective cancer vaccines.

Results Local and systemic antitumor response in tumor-bearing mice treated with tumor cells co-expressing IFN-a and tk Figure 1 shows the representative results of a set of experiments in which we evaluated the effectiveness of the genetically modified tumor vaccines co-expressing IFN-a and tk when both treatments (ie cells and GCV) were performed either homolaterally or contralaterally in mice bearing established FLC tumors. A single peritumoral injection of 107 IFNtk cells followed by GCV administration resulted in the complete tumor eradication and cure of the animals (Figure 1a). Interestingly, a marked inhibition of parental tumor growth and a significant increase in survival time were also observed when tumor-bearing mice were treated contralaterally with the same regimen (Figure 1b). When tumor-bearing mice were subjected to repeated contralateral injections of tk cells followed by GCV administration, there was an increase in the antitumor response with respect to mice receiving a single cell inoculation, and 10% of the animals proved to be cured (Figure 1c). Notably, repeated contralateral injections of the same numbers (106) of control tk cells (nonproducing IFN) followed by GCV treatments did not exhibit any therapeutic effect (data not shown). Only a slight antitumor response was observed in mice injected with IFN-a-tk cells but not treated with GCV (Figure 1c). To evaluate the possible effectiveness of a local treatment with cells co-expressing IFN-a and tk in animals with visceral metastases, we determined whether the combined therapy was also effective in mice preinjected intravenously (i.v.) with 500 (approximately 50 LD50) metastatic FLC. As shown in Figure 2, repeated s.c. injections of IFN-a-tk cells and GCV resulted in a marked increase in survival time as compared with control mice,

while only a slight antitumor response was observed when the same treatments were performed with control tk cells not producing IFN-a.

Analysis of the host cells involved in the antitumor response of tumor-bearing mice to the ‘IFN-a-tk/GCV’combined regimen We had previously shown that the ‘IFN-a-tk/GCV regimen’ did not induce any antitumor effect in splenectomized, irradiated, and anti-asialo-GM1-treated nude mice.7 These data indicated that the mechanisms underlying the therapeutic response to combined therapy were host-mediated. It was of interest, however, to characterize in detail which host mechanisms could be responsible for the impressive destruction of the parental tumor after GCV treatment and in particular which host cells could be involved in the generation of the local and systemic antitumor effects. Thus, we performed selective in vivo immunosuppression treatments by using monoclonal antibodies capable of depleting either specific T cell subsets (CD4 + or CD8+ cells) or polymorphonuclear cells, in tumor-bearing mice subsequently subjected to the combined ‘IFN-a-tk/GCV regimen’. The results of some representative experiments are shown in Table 1. Depletion of either CD4+ or CD8 + T lymphocytes resulted in the complete abrogation of the therapeutic effect (Table 1, experiments 1 and 3). Treatments with the antigranulocyte antibody resulted in some inhibition of the antitumor response (Table 1, experiments 1 and 2), thus suggesting that polymorphonuclear cells were also involved in the IFN-a-tk/GCV-induced destruction of the parental tumor. We had previously described the dramatic tumor cell degeneration occurring a few days after GCV treatment in tumor-bearing mice subjected to peritumoral injection with tk cells producing IFN.7 In this study, we have characterized the host cell infiltrates and the tumor cell morphology in tumor-bearing mice after the peritumoral or contralateral treatments. Immunohistochemical analysis of the tumor area in mice receiving the peritumoral injection of IFN-tk-expressing tumor cells and GCV showed the presence of several CD4+ and CD8+ T lymphocytes (Figure 3b and c); some granulocytes were found among the residual, severely damaged tumor cells and particularly at the periphery of necrotic areas (Figure 3d). Interestingly, morphological analysis of the parental tumor area after contralateral injection of tk-expressing tumor cells showed that the tumor mass was formed by aggregates of tumor cells interspersed with large areas of coagulative necrosis (Figure 4). Macrophages and CD4+ or CD8+ T lymphocytes were randomly scattered among the tumor cells (Figure 5b–d), and in some cases in close contact with tumor cells (Figure 5e). Notably, numerous apoptotic tumor cells were observed in the parental tumor area, as previously shown after peritumoral ‘Cl11tk528/GCV’ treatment.7 Comparison of the effectiveness of repeated injections of irradiated versus viable IFN-a/tk cells To explore the spectrum of the possible clinical applications of the combined ‘IFN-a-tk/GCV’ therapeutic strategy further, it was important to determine whether the combination of IFN-a and tk genes was also advantageous when irradiated (instead of viable) cells were used. In fact, a major concern about the use of viable IFN-

1247


1248

Figure 1 Effectiveness of tumor vaccines co-expressing IFN-a and the tk gene when injected either homolaterally or contralaterally in tumor-bearing mice. DBA/2 mice, 6–7 weeks old, were injected s.c. with 2 × 106 3Cl-8 parental tumor cells. One group of mice was used as a control for 3Cl-8 parental tumor growth and did not receive any further treatment (P). (a) After 3 days, one group of tumor-bearing mice was inoculated peritumorally with 107 Cl-11-IFNtk528 cells (g). Four days later, GCV (150 mg/kg) was administered at the site of tumor growth, twice a day for 5 days. The same schedule of GCV treatment was repeated after a week interval. There were six mice per group. (b) Three days later, one group of tumor-bearing mice was inoculated contralaterally with 107 Cl-11IFNtk528 cells (g) followed by GCV administration as described before. There were six mice per group. P , 0.05 versus 3Cl-8 injected mice. (c) Two groups of mice bearing 3-day tumors received repeated contralateral injections of 106 Cl-11-IFNtk528 cells (once a week for 4 weeks). In the first group, consisting of 10 mice, each injection was followed by one cycle of GCV administration (for 4 days, twice a day). The last injection was followed by two 5-day cycles with a week interval (g). P , 0.001 versus 3Cl-8 injected mice. In the second group, consisting of six mice, GCV was not administered to the vaccine-treated animals (G).

a-tk-expressing tumor cells derived from our previous observation that viable tk tumor cells were not eliminated by GCV in immunosuppressed nude mice.7 Moreover, in immunocompetent mice repeatedly injected with high numbers of viable tk cells, GCV did not succeed in totally suppressing tumor growth of the transduced cells. We then compared the effectiveness of repeated injections of

irradiated versus viable transduced cells, followed by GCV treatments, against the growth of established FLC or TS/A adenocarcinoma tumors. In the representative experiment shown in Figure 6a, treatment of mice bearing parental FLC tumors with repeated peritumoral injections of 107 irradiated IFN-a/tk-FLC and GCV caused complete tumor rejection and cure in 70% of the


the best antitumor response was clearly observed in mice also treated with GCV. Only a slight therapeutic effect was observed after repeated injections of irradiated control tk cells (not producing IFN) and GCV, indicating that IFN-a secretion was essential for in vivo augmentation of the tk-GCV-induced tumor killing.

Discussion

Figure 2 Effects of tumor vaccines co-expressing IFN-a and the tk in mice previously injected i.v. with parental metastatic FLC. Male DBA/2 mice, 6–7 weeks old, were injected i.v. with 500 (approximately 50 LD 50) 3Cl8 parental tumor cells. On day 1, the mice were divided into three groups. One group of six mice was used as a control and did not receive any further treatment (group A, P). In the other two groups, 10 mice per group received repeated s.c. injections of 2 × 106 3Cl-8tk122 (group B, G) or Cl-11-IFNtk528 (group C, g) (once a week for 4 weeks). Each injection was followed by one 4-day cycle of GCV administration. The last injection was followed by two 5-day cycles with a week interval. The P value of groups B and C versus group A was , 0.001. The P value of group C versus group B was , 0.01.

treated animals. On the contrary, when viable IFN-a/tk cells were used according to the same schedule, there was only an increase in survival time, but no cure was observed. (Under these conditions, the lack of a definite therapeutic response was due to the failure of GCV to eliminate the transduced viable cells, which were repeatedly injected.) Very similar results were obtained in mice bearing established poorly immunogenic TS/A tumors and receiving repeated injections of high doses of irradiated IFN-a/tk-TS/A cells and GCV (Figure 6b). Notably, in both tumor systems, the repeated injections of high numbers of irradiated IFN-a/tk cells proved rather effective even without GCV administration, but

Genetic transfer of the HSVtk into tumor cells followed by GCV administration has recently received increasing attention as a strategy for selective elimination of cancer cells. Since its first description,9 the HSVtk/GCVmediated tumor killing has been described in several cancer models including sarcoma,10 melanoma,11,12 colon carcinoma liver metastases13 and brain tumors.14–16 The initial studies showed that the ex vivo or in vivo HSVtk gene transfer and subsequent GCV treatments were very effective in determining regression of local tumor deposits both in peripheral tissues10,17,18 and in the brain.14–16 In particular, the efficacy of this approach was found to be superior to the expectations based on the relative proportion of tk-positive and tk-negative cells present in the tumor mass. This phenomenon has been termed the ‘bystander effect’8 and is considered to be the result of a number of different mechanisms, like the release of the toxic phosphorylated GCV by the tumor cells expressing the HSVtk following GCV-induced killing and its uptake by adjacent unmodified tumor cells via gap junctions or phagocytosis of apoptotic vesicles.8,19,20 Tumor necrosis consequent to the in vivo tk gene transduction of endothelial cells proliferating at the site of tumor growth and their killing after GCV administration has also been indicated as a mechanism potentiating the ‘bystander effect’.15 In these early studies, the participation of the host immune system in the ‘bystander effect’, as well as its importance in the development and maintenance of systemic antitumor immunity was either not taken into consideration or understated. In fact, initial observations such as the inability of dexamethasone treatment to diminish the effectiveness of the HSVtk/GCV

Table 1 Effects of selective immunosuppressive treatments on the antitumor response of mice to the combined ‘IFN-a/tk/GCV’ regimen Therapy (IFN-a/tk/GCV)

Immunosuppression treatment

No. of tumor-free mice a/Total No. of mice

Experiment 1

No Yes Yes Yes Yes

None None anti-CD4 anti-CD8 anti-PMN

0/6 3/6 0/6 0/6 1/6

Experiment 2

No Yes Yes

None None anti-PMN

0/5 9/9 3/6

Experiment 3

No Yes Yes Yes

None None anti-CD4 anti-CD8

0/6 5/6 0/6 0/6

DBA/2 mice, 6–7 weeks old, were inoculated s.c. with 2 × 106 3Cl-8 cells on day 0 and were left untreated or received a peritumoral injection of 107 Cl-11-IFNtk528 cells on day 3, followed by GCV administration starting on day 7, as described in Materials and methods. Antigranulocyte, anti-CD4 or anti-CD8 antibodies were administered i.v. on days 2, 3, 7, 11, 16, 21 and 25. a The reported values refer to mice remaining tumor-free when complete rejection of the primary tumor occurred in the group of mice subjected to the combined therapy and not receiving any immunosuppressive treatment (approximately day 40).

1249


1250

Figure 3 Histological and immunohistochemical features of the tumor area in a mouse bearing a 3-day 3Cl-8 parental tumor, subjected to the peritumoral ‘IFN-a/tk/GCV’ combined regimen. The Figure shows the pattern observed 7 days after injection of IFN-tk-expressing cells and 3 days since the beginning of GCV administration. (a) Residual tumor cells (arrowheads) are scattered in a fibroadipous tissue with foci of necrosis and several reactive cells; (b, c, d) cryostat sections tested with anti-CD8 (b), anti-CD4 (c), antigranulocyte (d) mAbs, showing several CD8+ and CD4+ T lymphocytes and some granulocytes (arrowheads) in the tumor area. (Magnification a–d, × 630).

Figure 4 Histological and ultrastructural features of the parental tumor area after contralateral injection of IFN-producing-tk-expressing tumor cells and GCV treatment. (a) At the edges of the large necrotic area (n), aggregates of severely damaged tumor cells can be observed (× 630); (b) ultrastructural detail of necrotic area constituted by amorphous material, disaggregated cell cytoplasms and naked cell nuclei (arrowheads) (× 3100).


1251

Figure 5 Histological, immunohistochemical and ultrastructural features of the parental 3Cl-8 tumor in a mouse subjected to contralateral ‘IFNa/tk/GCV’ regimen. The Figure shows the pattern observed 18 days after the beginning of combined therapy. (a) Residual aggregates of damaged tumor cells with some reactive cells at the periphery of a large necrotic area (n) (× 630); (b, c) cryostat sections tested with anti-CD4 (b) or anti-CD8 (c) mAbs, showing several CD4+ or CD8+ T lymphocytes in the tumor area (× 630); (d) ultrastructural micrograph showing an area at the periphery of the tumor mass. Amorphous necrotic material is present among residual damaged tumor cells and reactive cells as macrophages (arrows) and lymphocytes (arrowheads) (× 1900); (e) ultrastructural detail of a necrotic area border showing two lymphocytes (arrowheads) close to a tumor cell with a severely damaged cytoplasm (× 3800).

treatments in tumor-bearing rats, 21 or the failure of rats implanted with tk-expressing brain tumor cells and treated with GCV to reject wild-type tumors growing in the contralateral hemisphere16 led to the conclusion that the host immune system was not significantly involved in tk/GCV-mediated tumor regression. Subsequent studies, however, pointed out the limited and variable efficacy of the HSV/tk-GCV autotoxic system, depending on several factors, including efficiency of transduction,15 immunologic characteristics of different tumors as well as the state of the tumor cells with respect to the phase of the cell cycle during GCV administration, 11 and extent of gap junctions.22 Moreover, some of these studies indicated that immune components participated in the ‘bystander

effect’,13 whose magnitude was dramatically reduced in immunodeficient athymic mice,23 and that a tumor-specific protective immune response developed in animals subjected to the prodrug-induced killing of suicide genemodified tumor cells.24,25 In particular, the study by Barba and colleagues25 emphasized the crucial importance of antitumor immune mechanisms in suppressing the outgrowth of residual tumor cells after tk/GCV treatments. However, the development of a protective immune response occurred incidentally (only in some of the treated animals), suggesting that additional mechanisms, secondary to the tk/GCV tumor cell killing, must occur in order to sustain long-term tumor regression. From these results, it could be inferred that a more potent and long-


1252

Figure 6 Comparison of the effectiveness of several injections of irradiated versus viable tumor vaccines co-expressing IFN-a and the tk gene in tumor bearing mice. Male DBA/2 or female BALB/c mice, 6–7 weeks old, were injected s.c. with 2 × 106 3Cl-8 or 1 × 105 TS/A parental tumor cells, respectively. After 3 days, one group of mice received repeated peritumoral injections of 107 Cl-11-IFNtk528 or 106 I/A.4.60tk cells (once a week for 4 weeks), either irradiated (l) or viable (g). Each injection was followed by a 4-day cycle of GCV administration. The last injection was followed by two 5-day cycles with a week interval. The same schedule was used for the group injected with irradiated 3Cl-8tk122 or I/A.TCtk cells (G). Another group of mice was injected with irradiated Cl-11-IFNtk528 or I/A.4.60tk cells but did not receive GCV (K). One group of mice was used as a control for 3Cl-8 or TS/A parental tumor growth and did not receive any further treatment (P). There were six mice per group.

term antitumor response to the tk/GCV regimen could be achieved by combining this approach with cytokinegene therapy, since some immunoregulatory cytokines are capable of converting a local response into a systemic antitumor immune response, particularly important for those tumors showing a poor bystander effect and especially for metastatic cancers. Very little information is available on the therapeutic efficacy of the combination of cytokine genes and cytotoxic genes and apparently conflicting results have been obtained, depending on the tumor model or the site of tumor injection. In one case, the in situ transduction of previously implanted rat brain tumors with retroviral vectors expressing both interleukin-2 (IL-2) and tk genes did not result in any enhancement of tumor eradication compared with the transduction with the vector carrying the tk gene alone.21 However, IL-2 transduction of the same glioma cells implanted s.c. caused a potent inhibition of their growth, suggesting that the brain localization of the tumor was an obstacle to the development of an immune antitumor response.21 A more recent report26 showed that intratumoral administration in established hepatic metastases of adenoviral vectors carrying the HSVtk and mouse IL-2 genes, followed by GCV treatments, resulted in a significant antitumor response; however, no definite cure of the injected animals was observed. A similar short-term antitumor response was observed in a murine model for oral cancer.27 Very recently, using a different murine model for hepatic metastases of colon carcinoma, the same authors described how the combination of adeno-tk and adeno-IL-2 with a third adenoviral vector carrying the mouse granulocyte– macrophage colony-stimulating factor (mGM-CSF) potentiated the antitumor immunity and resulted in the long-term survival of a significant proportion of the treated animals.28 We have recently described the first example of an effective therapy of tumor-bearing mice by means of genetically modified tumor vaccines coexpressing both an

autotoxic gene (tk) and a cytokine gene (ie IFN-a1).7 In the present study, we have investigated the mechanisms of the potent antitumor response generated after this combined therapy. We have shown that repeated injections of tumor cells expressing tk and producing IFN-a followed by GCV treatments are effective in inhibiting subcutaneous tumors growing at distant sites as well as visceral metastases. Our data indicate that the generation of an effective systemic antitumor response strictly requires both the IFN-producing phenotype of the tk cells and GCV administration. Notably, the morphological and immunohistochemical analysis of the parental tumor area revealed zones of ischemic necrosis and the presence of infiltrating CD4+ and CD8+ T lymphocytes and macrophages in mice subjected to contralateral injections of IFN-a-tk-expressing tumor cells and GCV, similar to what was observed after peritumoral treatment. This observation further supports the notion that the host immune response elicited locally by the combined therapy is nevertheless capable of acting systemically against distantly growing tumors. Interestingly, the RT-PCR analysis of the cytokine expression at the site of tumor growth indicated that a consistent increase in the level of IFN-g mRNA, among all the cytokine mRNAs analyzed (TNF-a, TNF-b, IL-1b, IL-2, IL-10, IL-4, IL-12, IFN-g), occurred after the peritumoral injection of Cl-11tk528 cells (data not shown). The main role played by both CD4+ and CD8+ T lymphocytes in determining the successful outcome of the ‘IFN-a/tk/GCV’ therapy is clearly demonstrated by the lack of response to the combined therapy in mice depleted of these lymphocyte populations. Thus, taking into consideration the data presented in this article as well as in previous studies,6,7 the following kinetics of events can be envisaged: (1) the injection of IFN-a-producing tumor cells into tumorbearing mice may rapidly cause the recruitment of a prominent infiltrate of macrophages, neutrophils and lymphocytes within the parental tumor tissue as well as the appearance of initial areas of tumor cell degeneration,


possibly caused by secondary factors released by the infiltrating cells; (2) these early events are likely to create the conditions (through the recruitment of antigenpresenting cells, activation of some immune effector cells and continuous production of cytokines) for the generation of an initial antitumor immune response; (3) after GCV treatment, the massive tumor cell destruction may induce further functional interactions between tumor antigens and host reactive cells, thus resulting in the conversion of a local reaction into a systemic response, which is definitively mediated by immune T lymphocytes. In this article, we have also compared the effectiveness of irradiated versus viable tumor vaccines co-expressing the two genes. This issue is of particular importance, especially because of the concerns regarding the risk of using viable tumor vaccines in cancer patients. In fact, the use of any viable tumor vaccine expressing an autotoxic gene still poses the fundamental issue of the safety of the transduced tumor cells, which could persist after prodrug administration, especially when injected repeatedly in high numbers and in patients whose immune system may be somehow impaired. Our results clearly indicate that: (1) repeated injections of high doses of irradiated IFNa-tk-expressing tumor cells are more effective and safer than their viable counterparts for the treatment of mice bearing established metastatic tumors; (2) both IFNa production by the tumor cells and their cell death induced by GCV are necessary prerequisites for achieving an optimal antitumor response. These results were obtained using two unrelated tumors (ie FLC and TS/A), which show marked differences in their origin, pattern of in vivo growth and immunogenicity. In particular, it is worth mentioning that these tumors also exhibit marked differences in their response to the tk-GCVinduced tumor killing per se. In fact, while a single peritumoral injection of tk-expressing cells (not producing IFN) followed by GCV treatment only resulted in a slight antitumor response in the FLC system,7 the same treatment regimen induced a remarkable tumor rejection in the TS/A model (our unpublished results). Nevertheless, in both models, an optimal antitumor response was always found when tumor-bearing mice were injected with tumor cells co-expressing the two genes (IFN-a and tk). FLC and TS/A represent aggressive tumor models and 3-day-old 3Cl-8 or TS/A s.c. tumors can be considered established tumors on the basis of previous observations.29–32 Nevertheless, further experiments are needed to evaluate the therapeutic potential of the combined regimen against more advanced primary tumors or metastases, by increasing the dose or the number of injections of the irradiated vaccine. These experimentswill help to define the relevance of the ‘IFNa/tk/GCV’ combination therapy to the treatment of human malignancies. The observation that repeated treatments with irradiated IFN-a-tk-expressing tumor cells followed by GCV administration result in the cure of the vast majority of mice bearing established metastatic tumors strongly supports the possibility of using irradiated tumor cells expressing both a cytokine and an autotoxic gene for the development of more effective and safe cancer vaccines. Recently, the clinical use of allogeneic irradiated tk-positive ovarian cancer cells has been proposed for the treatment of patients with ovarian cancer.33 The rationale of this clinical protocol was based on previous results

indicating that the injection of irradiated HSVtk genemodified human colon carcinoma cells followed by GCV treatments could prolong the survival of mice bearing intraperitoneal syngeneic fibrosarcomas.34 The authors concluded that irradiated xenogeneic tk-expressing tumor cells could generate a potent ‘bystander effect’ in vivo, enhanced by immunostimulation caused by the release of inflammatory cytokines. 34 Regarding the perspectives of further clinical studies, our results indicate that a direct combination of a cytokine gene with an autotoxic gene in a tumor vaccine could result in a much more impressive antitumor response and in the development of both local and systemic antitumor immunity. Several clinical trials with genetically modified tumor cells producing various cytokines are currently under way. Recently, however, the initial enthusiasm which accompanied the start of these clinical studies has been replaced by the focus of attention on the development of new and more effective strategies in the preparation of cancer vaccines. Although we do not know the potential effectiveness of combining other cytokine genes with the HSVtk/GCV system, IFN-a is a good candidate for this type of tumor vaccine strategy. IFN-a was the first cytokine to be used clinically for the treatment of some malignancies and extensive information is now available on its role in the generation and action of immune T lymphocytes.35 The importance of IFN-a for the generation of a protective antitumor immunity has been demonstrated in several experimental models.35 It may be worth noting that some human cancers in which IFN-a is somehow effective (ie melanoma, ovarian carcinoma, renal cell cancer) are those in which the development of an antitumor immune response is considered to be important. Nevertheless, the use of IFN-a in cancer gene therapy protocols has not yet been taken into consideration. We thus conclude that: (1) IFN-a is an interesting cytokine to be considered in cytokine gene therapy of cancer; (2) vaccines co-expressing an autotoxic gene and a cytokine gene may represent promising new tools for the treatment of some cancer patients.

Materials and methods Mice Male DBA/2 mice, 6–7 weeks old were purchased from Charles River Breeding Laboratories (Italia Calco, Italy). Tumor cells 3Cl-8, an IFN-a/b-resistant clone of FLC passaged in vitro, was originally obtained from Dr E Affabris.36 The cells were subsequently passaged in vivo by weekly i.p. injections into DBA/2 mice. These in vivo passaged 3Cl8 FLC were highly metastatic for the liver and the spleen.29 TS/A is a highly aggressive and metastasizing cell line established from the first in vivo transplant of a moderately differentiated mammary adenocarcinoma that arose spontaneously in a 20-month-old multiparous BALB/c mouse.30 The minimal 100% TS/A parental cells (TS/A-pc) tumor-inducing dose is 4 × 104 cells in syngeneic BALB/c mice. The Cl-11tk528 and the I/A.4.60tk clone were obtained after transfection of the IFN-producing FLC clone IFN-a1-Cl-113 and TS/A clone I/A.4.60,4 respectively, with the tgCMV/HyTK plasmid containing a hygromycin phosphotransferase-thymidine kinase

1253


1254

fusion gene as previously described.7 Cl-11tk528 and I/A.4.60tk cells secreted about 500 and 60 IU/ml of IFNa, respectively. The 3Cl-8tk122 and TS/Atk clones, obtained after transfection of 3Cl-8 parental FLC and TS/A-pc with the tgCMV/HyTK plasmid, were used as controls for the potential effects of vector sequences. The susceptibility to the cytotoxic action of GCV in vitro as well as in vivo was previously demonstrated for both Cl11tk528 and 3Cl-8tk122 cells7 and was subsequently also assessed for I/A.4.60tk and TS/Atk cells (unpublished results).

Combined therapy schedules For the therapy experiments, we defined an established tumor as a 3-day-old tumor formed after s.c. injection of 2 × 106 3Cl-8 or 1 × 105 TS/A parental tumor cells. This definition is justified by the rapid in vivo growth after s.c. injection and by the histologic features of these tumors grown in syngeneic mice.29–32 For evaluation of local effect, mice bearing 3-day-old FLC tumors were treated with a single peritumoral injection of 107 Cl-11tk528 cells. Four days later, GCV (150 mg/kg) was administered twice a day at the site of tumor growth for two cycles of 5 days, the second following the first by a week. Different treatment schedules for the evaluation of systemic effect were applied as described in detail in the Figure legends. In some experiments, tk-expressing FLC or TS/A tumor cells were irradiated with 5000 rad. After irradiation, Cl11tk528 and I/A.4.60tk cells were reseeded in triplicate and culture supernatants were collected every 24 h for 5 days to titrate the amount of biologically active secreted IFN, as previously described.37 Considerable levels of IFN (256–512 IU/ml in Cl-11tk528 cell culture supernatants and 64–128 IU/ml in I/A.4.60tk ones) were detected up to 5 days after irradiation. Sensitivity to in vitro cytotoxic action of GCV of viable and irradiated tkexpressing tumor cells was determined as previously described.7 In vivo immunosuppression treatments The antibodies used to deplete specific cell populations were purified by ammonium sulfate precipitation from ascites derived from clone GK1.5 (antiCD4), clone TIB 105 (antiCD8), and clone RB6–8C5 (antigranulocyte). Details on antibody preparations and injection schedules are described elsewhere.38 Morphological analysis For histologic evaluation, tissue samples were fixed in 10% neutral buffered formalin, embedded in paraffin, sectioned at 4 mm, and stained with hematoxylin–eosin or Giemsa. For electron microscopy, specimens were fixed in cacodylate buffered 2.5% glutaraldehyde, postfixed in osmium tetroxide, and then embedded in Epon 812 (Electron Microscopy Science, Fort Washington, PA, USA). Ultrathin sections were stained with uranyl acetate–lead citrate. For immunohistochemistry, acetonefixed cryostat sections were incubated for 30 min with anti-L3T4 (CD4), anti-Lyt-2 (CD8a) (all from Sera-Lab, Crawley Down, UK), anti-MAC-1a (CD11b/CD18), antiMAC-3 (all from Boehringer Mannheim, Milan, Italy), and antigranulocyte (clone RB6–8C5) rat monoclonal antibodies. After washing, they were overlaid with biotinylated rabbit antirat immunoglobulins (Ig) (Vector Laboratories, Burlingame, CA, USA) for 30 min. Unbound Ig

were removed by washing, and the slides were incubated with ABC complex/AP (Dako, Glostrup, Denmark).

Statistical analysis Data were analyzed by Wilcoxon’s sum rank test.

Acknowledgements We are indebted to Mrs Anna Ferrigno for her valuable secretarial assistance. This work was supported in part by the Associazione Italiana Ricerca sul Cancro, and the Gene Therapy Program, Ministry of Health, Italy.

References 1 Tepper RI, Pattengale PK, Leder P. Murine interleukin-4 displays potent antitumor activity in vivo. Cell 1989; 57: 503–512. 2 Hussein AM. The potential applications of gene transfer in the treatment of patients with cancer: a concise review. Cancer Invest 1996; 14: 343–352. 3 Ferrantini M et al. a1-Interferon gene transfer into metastatic Friend leukemia cells abrogated tumorigenicity in immunocompetent mice: antitumor therapy by means of interferon-producing cells. Cancer Res 1993; 53: 1107–1112. 4 Ferrantini M et al. IFN-a1 gene expression into a metastatic murine adenocarcinoma (TS/A) results in CD8+ T cell-mediated tumor rejection and development of antitumor immunity. Comparative studies with IFN-g-producing TS/A cells. J Immunol 1994; 153: 4604–4615. 5 Kaido T et al. IFN-a1 gene transfection completely abolishes the tumorigenicity of murine B16 melanoma cells in allogeneic DBA/2 mice and decreases their tumorigenicity in syngeneic C57BL/6 mice. Int J Cancer 1995; 60: 1–9. 6 Gabriele L et al. Local and systemic response of mice to interferon-a-1-transfected Friend leukemia cells. Am J Pathol 1995; 147: 445–460. 7 Santodonato L et al. Cure of mice with established metastatic Friend leukemia cell tumors by a combined therapy with tumor cells expressing both interferon-a1 and herpes simplex thymidine kinase followed by ganciclovir. Hum Gene Ther 1996; 7: 1–10. 8 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. 9 Moolten F. Tumor chemosensitivity conferred by inserted herpes thymidine kinase genes: paradigm for a prospective cancer control strategy. Cancer Res 1986; 46: 5276–5281. 10 Moolten FL, Wells JM. Curability of tumors bearing herpes thymidine kinase genes transferred by retroviral vectors. J Natl Cancer Inst 1990; 82: 297–300. 11 Golumbek PT et al. Herpes simplex-1 virus thymidine kinase gene is unable to completely eliminate live, nonimmunogenic tumor cell vaccines. J Immunother 1992; 12: 224–230. 12 Vile RG, Hart IR. Use of tissue-specific expression of the herpes simplex virus thymidine kinase gene to inhibit growth of established murine melanomas following direct intratumoral injection of DNA. Cancer Res 1993; 53: 3860–3864. 13 Caruso M et al. Regression of established macroscopic liver metastases after in situ transduction of a suicide gene. Proc Natl Acad Sci USA 1993; 90: 7024–7028. 14 Culver KW et al. In vivo gene transfer with retroviral vector– producer cells for treatment of experimental brain tumors. Science 1992; 256: 1550–1552. 15 Ram Z et al. In situ retrovirus-mediated gene transfer for the treatment of brain tumors in rats. Cancer Res 1993; 53: 83–88. 16 Barba D, Hardin J, Ray J, Gage FH. Thymidine kinase-mediated killing of rat brain tumors. J Neurosurg 1993; 79: 729–735. 17 Borrelli E, Heyman R, Hsi M, Evans RM. Targeting of an inducible toxic phenotype in animal cells. Proc Natl Acad Sci USA 1988; 85: 7572–7576.


18 Ezzedine ZD et al. Selective killing of glioma cells in culture and in vivo by retrovirus transfer of the herpes simplex virus thymidine kinase gene. New Biologist 1991; 3: 608–614. 19 Li Bi W, Parysek LM, Warnick R, Stambrook PJ. In vivo evidence that metabolic cooperation is responsible for the bystander effect observed with HSVtk retroviral gene therapy. Hum Gene Ther 1993; 4: 725–731. 20 Colombo BM et al. The ‘bystander effect’: association of U-87 cell death with ganciclovir-mediated apoptosis of nearby cells and lack of effect in athymic mice. Hum Gene Ther 1995; 6: 763–772. 21 Ram Z et al. In vivo transfer of the human interleukin-2 gene: negative tumoricidal results in experimental brain tumors. J Neurosurg 1994; 80: 535–540. 22 Chen S-H 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. 23 Vile RG et al. Systemic gene therapy of murine melanoma using tissue specific expression of the HSVtk gene involves an immune component. Cancer Res 1994; 54: 6228–6234. 24 Mullen CA, Coale MM, Lowe R, Blaese RM. Tumors expressing the cytosine deaminase suicide gene can be eliminated in vivo with 5-fluorocytosine and induce protective immunity to wildtype tumor. Cancer Res 1994; 54: 1503–1506. 25 Barba D, Hardin J, Sadelain M, Gage FH. Development of antitumor immunity following thymidine kinase-mediated killing of experimental brain tumors. Proc Natl Acad Sci USA 1994; 91: 4348–4352. 26 Chen S-H et al. Combination gene therapy for liver metastasis of colon carcinoma in vivo. Proc Natl Acad Sci USA 1995; 92: 2577–2581. 27 O’Malley BW Jr et al. Combination gene therapy for oral cancer in a murine model. Cancer Res 1996; 56: 1737–1741. 28 Chen S-H et al. Combination suicide and cytokine gene therapy for hepatic metastases of colon carcinoma: sustained antitumor immunity prolongs animal survival. Cancer Res 1996; 56: 3758– 3762.

29 Belardelli F et al. Biologic and biochemical differences between in vitro and in vivo passaged Friend erythroleukemia cells. I. Tumorigenicity and capacity to metastasize. Int J Cancer 1984; 30: 389–395. 30 Nanni P et al. TS/A: a new metastasizing cell line from a BALB/c spontaneous mammary adenocarcinoma. Clin Exp Metas 1983; 1: 373–380. 31 Belardelli F et al. Antitumor effects of interferon in mice injected with interferon-sensitive and interferon-resistant Friend leukemia cells. III. Inhibition of growth and necrosis of tumor implanted subcutaneously. Int J Cancer 1983; 31: 649–653. 32 Allione A et al. Immunizing and curative potential of replicating and nonreplicating murine mammary adenocarcinoma cells engineered with interleukin (IL)-2, IL-4, IL-6, IL-7, IL-10, tumor necrosis factor a, granulocyte-macrophage colony-stimulating factor, and g-interferon gene or admixed with conventional adjuvants. Cancer Res 1994; 54: 6022–6026. 33 Freeman SM et al. The treatment of ovarian cancer with a gene modified cancer vaccine: a phase I study. Hum Gene Ther 1995; 6: 927–939. 34 Freeman SM et al. The role of cytokines in mediating the bystander effect using HSVtk xenogenic cells. Cancer Lett 1995; 92: 167–174. 35 Belardelli F, Gresser I. The neglected role of type I interferon in the T cell response: implications for its clinical use. Immunol Today 1996; 8: 369–372. 36 Affabris E, Jemma C, Rossi GB. Isolation of interferon-resistant variants of Friend erythroleukemia cells: effects of interferon and ouabain. Virology 1982; 120: 441–452. 37 Belardelli F et al. Studies on the expression of spontaneous and induced interferons in mouse peritoneal macrophages by means of monoclonal antibodies to mouse interferons. J Gen Virol 1987; 68: 2203–2212. 38 Ciolli V et al. Combined IL-1/IL-2 therapy of mice injected with highly metastatic Friend leukemia cells: host antitumor mechanisms and marked effects on established metastases. J Exp Med 1991; 173: 313–322.

1255