True

Gene Therapy (2002) 9, 867–878  2002 Nature Publishing Group All rights reserved 0969-7128/02 $25.00 www.nature.com/gt

RESEARCH ARTICLE

Differential effects of angiostatin, endostatin and interferon-␣1 gene transfer on in vivo growth of human breast cancer cells S Indraccolo1,2, E Gola2, A Rosato2, S Minuzzo2, W Habeler2, V Tisato2, V Roni2, G Esposito2, M Morini3, A Albini4, DM Noonan3, M Ferrantini5, A Amadori2 and L Chieco-Bianchi2 1

IST-Viral and Molecular Oncology Section-Padova, Padova, Italy; 2Department of Oncology and Surgical Sciences, University of Padova, Padova, Italy; 3IST-Modulo di Progressione Neoplastica, Genova, Italy; 4IST-Laboratorio di Biologia Molecolare, Genova, Italy; and 5Laboratory of Virology, Istituto Superiore di Sanita`, Roma, Italy

The administration of different angiogenesis inhibitors by gene transfer has been shown to result in inhibition of tumor growth in animal tumor models, but the potency of these genes has been only partially evaluated in comparative studies to date. To identify the most effective anti-angiogenic molecule for delivery by retroviral vectors, we investigated the effects of angiostatin, endostatin and interferon(IFN)-␣1 gene transfer in in vivo models of breast cancer induced neovascularization and tumor growth. Moloney leukemia virus-based retroviral vectors for expression of murine angiostatin, endostatin and IFN-␣1 were generated, characterized, and used to transduce human breast cancer cell lines (MCF7 and MDA-MB435). Secretion of the recombinant proteins was confirmed by biological and Western blotting assays. Their production did not impair in vitro growth of these breast cancer cells nor their viability, and did not

interfere with the expression of angiogenic factors. However, primary endothelial cell proliferation and migration in vitro were inhibited by supernatants of the transduced cells containing angiostatin, endostatin, and IFN-␣1. Stable gene transfer of the IFN-␣1 cDNA by retroviral vectors in both MCF7 and MDA-MB435 cells resulted in a marked and longlasting inhibition of tumor growth in nude mice that was associated with reduced vascularization. Endostatin reduced the in vivo growth of MDA-MB435, but not MCF7 cells, despite similar levels of in vivo production, and angiostatin did not impair the in vivo growth of either cell line. These findings indicate heterogeneity in the therapeutic efficacy of angiostatic molecules delivered by viral vectors and suggest that gene therapy with IFN-␣1 and endostatin might be useful for treatment of breast cancer. Gene Therapy (2002) 9, 867–878. doi:10.1038/sj.gt.3301703

Keywords: angiostatin; endostatin; interferon; retroviral vectors; gene therapy

Introduction Anti-angiogenic therapy represents a promising approach to cancer treatment because most solid tumors cannot grow without the necessary supply of oxygen and nutrients ensured by the formation of new blood vessels. Moreover, metastatic spread of solid tumors depends on the vascularization of the primary mass, indicating that a blockage of tumor angiogenesis will also block tumor metastasis. A wide range of angiogenesis inhibitors has been tested, many of which showed efficacy against a variety of solid tumor types, in particular the endogenous inhibitors, such as angiostatin, endostatin, serpin antithrombin, the soluble form of the vascular endothelial growth factor (VEGF) receptor 1, and some members of the interferon (IFN) family (reviewed by Kong and Crystal1 and Cao2). These proteins are good candidates for gene delivery because they are produced physiologically, indicating that they are probably not immunogenic, and Correspondence: S Indraccolo, IST-Viral and Molecular Oncology Section and Department of Oncology and Surgical Sciences, University of Padova, via Gattamelata 64, 35128-Padova, Italy Received 20 September 2001; accepted 15 February 2002

effective extracellularly, which circumvents the need to transduce all tumor cells. Originally isolated from the serum of Lewis lung carcinoma-bearing mice, angiostatin is a 38-kDa internal fragment of plasminogen that spans the first four kringle domains generated by cleavage of plasminogen by a macrophage-derived metalloelastase or other matrix metalloproteinases.3 Endostatin is a 20-kDa COOH-terminal fragment of collagen XVIII.4 Both angiostatin and endostatin have been reported to inhibit specifically endothelial cell proliferation and migration in vitro. Furthermore, both inhibited experimental primary tumor growth as well as angiogenesis-dependent growth of metastases in mice, inducing tumor dormancy without detectable toxicity or development of resistance.5,6 IFN-␣ and IFN-␤ are multifunctional regulatory cytokines involved in the control of cell proliferation and viral infections that also show anti-angiogenic activity both in vitro and in vivo.7,8 Frequent systemic administration of low-dose IFN-␣ and IFN-␤, or the introduction of their genes in some tumor or stromal cells has produced therapeutic effects against many tumor types in several animal models.9–11 Angiostatic therapy appears to require long-term administration of the inhibitor to ensure tumor growth

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

868

suppression in vivo.1 Long-term systemic delivery of recombinant molecules is expensive, time-consuming for the patient, and may be insufficient to obtain high local concentrations of the therapeutic molecule in the tumor mass. The delivery of the molecule through a gene therapy approach could be a solution, as it would lead to constantly high local levels of the anti-angiogenic protein. Although a number of studies address the therapeutic activity of angiogenesis inhibitors in tumor models, few compare their efficacy when administered by gene transfer.12 The identification of an ‘optimal’ candidate for gene therapy studies clearly involves many steps, among others a definition of the level required for therapeutic effects, and in this regard, some marked differences among the angiostatic molecules might exist. Indeed, it is clear that anti-angiogenic therapy with recombinant angiostatin and endostatin demands a high dose of proteins, in the range of 10–20 mg/kg/day in mice, but it is largely unknown what levels should be reached within the tumor to observe a therapeutic effect. On the other hand, IFN-␣ seems more active as an angiostatic drug when administered systemically at relatively low doses.9 The issue of dose–response effects is particularly relevant to gene therapy approaches, because the available viral and non-viral gene transfer systems have a limited capability to express the transgene in vivo. Therefore, we studied which angiogenesis inhibitor might be the most effective when delivered by retroviral vectors, one of the few approved vector systems that could confer long-term expression of angiostatic molecules in patients. We generated MLV-based retroviral vectors for murine angiostatin, endostatin and IFN-␣1, and used the vectors to deliver these genes to two different breast cancer cell lines in parallel. We observed that stable gene transfer of IFN-␣1 and endostatin cDNAs, but not angiostatin cDNAs resulted in delayed tumor growth in nude mice. These findings confirm that endogenous inhibitors of angiogenesis administered by a gene therapy approach could be useful for cancer treatment, and clearly indicate that the therapeutic effects of these agents delivered by viral vectors depend on the tumor type.

Results Generation and characterization of retroviral vectors expressing angiogenesis inhibitors The Moloney murine leukemia virus (Mo-MLV)-based MFG retroviral vector served as the basic vector to derive the murine angiostatin- and endostatin-expressing vectors used in this study (Figure 1a). A retroviral vector expressing the green fluorescent protein as the reporter gene, termed LE, was used to evaluate the efficiency of gene delivery to the breast cancer cells. The IFN-␣1expression vector, termed LMuIFN␣1SN, was previously described by others and is based on the LXSN backbone (Figure 1a).13 Retroviral vector-containing supernatants were generated by transient transfections of 293T packaging cells, and used to transduce both MCF7 and MDAMB435 breast cancer cells, and fibroblastic NIH-3T3 cells. As some of the vectors used did not contain a selectable marker, gene transfer was repeated several times to maximize the fraction of transduced cells. The percentage of genetically modified cells was estimated by measuring enhanced green fluorescent protein (EGFP) expression in Gene Therapy

Figure 1 Angiostatin and endostatin expression vector design and in vitro expression studies. (a) Schematic representation of the constructs used in this study: details of molecular cloning are reported in the Materials and methods. The HA tag was fused to the C-terminus of the angiostatin fragment. The human B29 leader sequence was fused to the N-terminus of endostatin cDNA to allow its extracellular secretion. SD and SA indicate the splice donor and the splice acceptor sites of the MFG retroviral vector; ⌿ is the packaging signal of the vector. Restriction sites relevant to cloning are indicated. The SV40-Neo box in the LMuIFN␣1SN vector indicates the neomycin phosphotransferase gene driven by the SV40 promoter. (b) Expression of angiostatin and endostatin by retroviral vector-transduced MDA-MB435 and MCF7 cells as determined by Western blotting. In each panel, supernatants from MFG-mAST-, MFG-endo-, IFN-␣1-, and control LE-transduced cells are analyzed. Murine HAtagged angiostatin generated a band of 58 kDa; endostatin and IFN-␣1 generated bands of 20 kDA.

the cells transduced by the LE vector in parallel experiments, and was >90% for each cell line (data not shown). Angiogenesis inhibitor production by the genetically modified cells was subsequently analyzed by immunoblotting analysis of conditioned medium from the transduced cells (Figure 1b). Both MCF7 and MDA-MB435 cells released detectable amounts of angiogenesis inhibitors into the culture medium. Immunoblotting of supernatants indicated that the MFG-mAST vector conferred angiostatin expression at relatively high-levels in vitro (Figure 1b). Its apparent molecular weight was about 58 kDa, which is higher than that reported for murine angiostatin,3 partly because of the presence of an HA tag attached to the C-terminus as a fusion protein, as previously described.14 No immunoreactive bands were detected in the supernatants of the LE-transduced cells. Radioimmunoprecipitation analysis with the anti-HA mAb confirmed these findings (data not shown). Since a murine angiostatin-specific quantitative ELISA is not available, we estimated angiostatin levels by Western blotting, and found that transduced MCF7 and MDAMB435 cells generated similar levels of angiostatin (5– 10 ␮g/ml). Transduction of both MCF7 and MDA-MB435 cells by the MFG-endo retroviral vector resulted in the production of murine endostatin by the breast cancer cells, clearly visible as a 20 kDa band of similar intensity in the supernatants obtained from each cell line upon immuno-

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

blotting. The production of endostatin in vitro was quantified by an ELISA assay, which indicated concentrations of 325 ng/ml and 310 ng/ml in the 24 h-conditioned medium of retroviral-vector transduced MCF7 and MDA-MB435 cells, respectively. Background levels of endostatin (⬍2 ng/ml) were detected in the supernatants of breast cancer cells transduced by the LE vector and in the parental cell supernatants. Repeated sampling of the supernatants showed that angiostatin and endostatin production by MCF7 and MDA-MB435 transduced cells was stable, with no apparent reduction even after 45 days in vitro culture (data not shown). Interferon production by the IFN-␣1 retroviral vector transduced cells was determined by Western blotting (Figure 1b) and quantified by an IFN titration assay, which indicated levels of 256–512 IU/ml in supernatants of both MCF7 and MDA-MB435 cells. In vitro growth kinetics of the genetically modified cells and analysis of basic fibroblast growth factor (bFGF) and VEGF expression We next determined whether the expression of the angiogenesis inhibitors, or perhaps the repeated cycles of gene transfer by retroviral vectors, might have affected the in vitro growth kinetics of the breast cancer cell lines. In vitro tumor cell growth impairment has never been reported for angiostatin and endostatin, while IFN-␣1 produced by IFN-␣1 retroviral vector genetically modified cells might have direct antiproliferative effects. The proliferation of wild-type MCF7 and MDA-MB435 cells in the presence of IFN-␣1-containing medium was identical to that of control cells (data not shown), suggesting that the former are not directly sensitive to IFNs and that IFN gene transfer would not lead to the selection of IFNresistant breast cancer cells. Proliferation assays demonstrated that neither angiostatin, endostatin nor interferon at the levels detected in this study, significantly affected the proliferation of breast cancer cells in vitro. The growth kinetics of the angiostatin, endostatin and IFN-␣1-producing cells were similar to those of cell lines transduced with the EGFP-encoding vector or the parental cells (Figure 2a). To determine whether gene delivery by MLV vectors may induce alterations in the cell cycle profiles of the target cells, we stained them with propidium iodide and analyzed the cell cycle by flow cytometry. We found that a significant fraction of parental MCF7 cells growing in vitro were cycling cells, as the G0/G1 and the S/G2 cell populations were 29% and 71%, respectively. Furthermore, analysis of the transduced cells disclosed patterns similar to those of the parental cells (data not shown), thus indicating that retroviral vector gene delivery did not grossly change the growth characteristics of the cells. Similar results were obtained with MDAMB435 cells. To investigate whether repeated transductions may affect cell viability, which in turn might influence tumor growth in vivo, we verified it by eosin staining and a twocolor fluorescence viability assay; in all cases >99% of the cells were alive, with no appreciable differences among the different samples. The results of a representative analysis on MCF7 cells are shown in Figure 2b. To evaluate whether the production of angiostatic molecules might interfere with the expression of angiogenic molecules by the tumor cell lines, and thus possibly with their capacity to form tumors in vivo, we evaluated

human bFGF and VEGF expression by RT-PCR. We observed that MCF7 cells expressed only VEGF in vitro, while MDA-MB435 cells expressed both VEGF and bFGF (Figure 2c). RT-PCR with VEGF-specific primers amplified two bands of 543 and 403 bp in all samples, corresponding to the VEGF121 and VEGF165 isoforms, respectively. On the other hand, RT-PCR with bFGF-specific primers amplified one band of 239 bp in all samples. No significant alteration in the amount of VEGF or bFGF PCR products was observed in the transduced human cell lines producing angiostatin, endostatin, or murine IFN-␣1, as compared with the levels detected in the control cells (LE) or parental cell lines (Figure 2c), suggesting no significant alterations in transcription of these mRNAs.

869

Effects of angiogenesis inhibitors produced by breast cancer cells on endothelial cells in vitro Filtered conditioned culture medium was collected from each cell line, and layered over P4-P8 passage human umbilical vein endothelial cells (HUVEC) to examine directly the effects on HUVEC proliferation using a [3H]thymidine incorporation assay. We observed that both IFN-␣1 and endostatin dramatically inhibited HUVEC proliferation, while angiostatin had a less pronounced effect (Figure 3a). HUVEC readily migrated in response to conditioned medium from NIH-3T3 cells in chemotaxis assays. When pre-incubated with supernatants of IFN-␣1 expressing MCF7 cells, an approximately 50% reduction in HUVEC migration was observed as compared with controls (Figure 3b). Angiostatin- and endostatin-containing supernatants did not significantly impair HUVEC migration under the same experimental conditions. However, when produced at higher levels by 293T cells transiently transfected by the same vectors, whose conditioned medium contained about five-fold more factor than conditioned medium from transduced MCF7 cells (data not shown), angiostatin and endostatin reduced endothelial cell migration by 60% and 47%, respectively (Figure 3c). In vivo tumor growth inhibition Genetically modified breast cancer cells were implanted s.c. in nude mice, and the tumors generated were measured over time. In the case of MCF7 cells, neither angiostatin nor endostatin significantly reduced tumor growth as compared with the EGFP-expressing retroviral vector transduced control cells (Figure 4a). In contrast, the IFN␣1-producing MCF7 cells showed much slower tumor growth than controls (Figure 4a), demonstrating a marked therapeutic effect of murine IFN. The tumors formed by MDA-MB435 cells were generally smaller than those formed by MCF7 cells, and grew slower in nude mice. Tumors formed by the different experimental groups over the first 6 weeks were similar in size, after which the tumors generated by endostatin- or IFN-␣1producing MDA-MB435 cells underwent a marked reduction in size as compared with controls (Figure 4b). Angiostatin did not significantly reduce tumor growth in this system (Figure 4b). The divergent effects of endostatin on the growth of these two breast cancer cell lines were not due to significant differences in the amount of endostatin produced in vitro, as determined in the conditioned medium by immunoblotting and confirmed by ELISA (see above). Gene Therapy

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

870

Figure 2 In vitro effects of the angiogenesis inhibitors on breast cancer cell proliferation, viability and bFGF or VEGF expression. (a) In vitro growth rates of parental (wt), angiostatin-, endostatin-, IFN-␣1-, or EGFP-producing MDA-MB435 (left panel) and MCF7 (right panel) cells as determined by [3H]-thymidine incorporation. No significant differences in proliferation rate were found among the different cell populations. (b) Flow cytometric viability assay using a cell viability/cytotoxicity kit (Molecular Probes). Parental or retroviral vector-transduced MCF7 cells were stained with calcein AM and ethidium homodimer-1 as detailed in the Materials and methods. After 5 min, flow cytometry analysis was carried out with excitation at 488 nm. The fluorescence emission was acquired separately at 530 nm and 585 nm for the live-cell (bottom panels) and the dead-cell (top panels) population, respectively. Ethanol-fixed MCF7 cells were used to define the dead cell fraction. (c) Expression of human bFGF and VEGF by the parental and the transduced cell lines in vitro by RT-PCR. MDA-MB435 cells express both angiogenic factors, while MCF7 express only VEGF in vitro. Angiogenesis inhibitor production did not interfere with bFGF and VEGF expression.

Histological studies The histology of the tumors formed by the different transduced MDA-MB435 cells after 14 days of growth in vivo was similar, with the tumor cells arranged in nests and solid patterns (Figure 5a–d). However, the tumors generated by the IFN-␣1-producing cells showed extensive areas of ischemic coagulative necrosis formed of cell debris, bordered by viable tumor cells with several mitotic figures (Figure 5d). Immunohistochemical analyses with anti-CD31 mAb clearly indicated a reduced vascularization in the IFNGene Therapy

␣1-producing tumors (Figure 5h) compared with controls (Figure 5e). No appreciable differences in the number of blood vessels were found between control MDA-LE and MDA-angio or MDA-endo cells (Figure 5e–g). Microvessel density counts in MDA-MB435 tumors indicated 132, 140, 131 and 84 microvessels in 10 representative fields of tumors formed by control MDA-LE, MDA-endo, MDAangio, and MDA-IFN-␣1 cells, respectively. Similar findings were obtained following histological and immunohistochemical analysis of the different MCF7 tumors (data not shown).

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

871

Figure 3 Effects of the angiogenesis inhibitors on the proliferation and migration of endothelial cells. (a) Proliferation of HUVEC following incubation with conditioned medium from endostatin-, IFN-␣1-, and, to a lesser extent, angiostatin-producing MCF7 cells, as compared with incubation with supernatants from MCF7 cells transduced with the control vector (LE) or DMEM (medium). The assay was performed in collagen I-coated 96-multiwell plates as detailed in the Materials and methods and repeated twice. Bars, s.d. (b) Effects of conditioned medium from transduced MCF7 cells (MCF7-CM) on chemotaxis of HUVEC. NIH-3T3conditioned medium was used as chemoattractant; serum-free medium (SFM) with 0.1% BSA was used as a negative control to evaluate random migration. To evaluate the effects of the different angiogenesis inhibitors on endothelial cell migration, HUVEC were pre-incubated with identical amounts of conditioned medium from MCF7 cells transduced with the different retroviral vectors, as detailed in Materials and methods. NIH3T3, migration to supernatant of NIH-3T3 cells; SFM, migration to serum-free medium; LE, mAST, endo, IFN-␣1 columns: migration of HUVEC to supernatant of NIH-3T3 cells following pre-incubation with conditioned medium from MCF7 cells transduced by the LE, angiostatin, endostatin, and IFN-␣1 retroviral vectors, respectively. The assay was performed in triplicate in Boyden chambers and repeated twice. Bars, s.d. (c) Effects of conditioned medium from transiently transfected 293T cells (293T-CM) on chemotaxis of HUVEC. The assay was performed as detailed in (b). Vectors and transfected cells are indicated as above.

Determination of in vivo expression of the angiogenesis inhibitors Different approaches were employed to confirm that genetically modified cells produced the angiogenesis inhibitors in vivo. Following the death of the animals, tumors

Figure 4 Patterns of in vivo growth of tumors by breast cancer cells producing angiogenesis inhibitors. MCF7 (panel a) and MDA-MB435 cells (panel b) were transduced ex vivo by retroviral vectors to express either angiostatin (open circles), endostatin (filled squares), IFN-␣1 (open squares) or EGFP (control vector) (filled circles). The genetically modified cells were subsequently mixed with matrigel and injected s.c. into the flanks of nude mice (six mice/group). Nodule dimensions were used to compute tumor volumes. Statistically significant differences between the experimental groups and controls (Student’s test) are indicated by asterisks. The experiment was repeated with another set of animals (six mice/group) with identical results.

were analyzed for expression of transgene-encoded mRNAs by RT-PCR. Tumor RNA was treated with DNase I to remove any residual genomic DNA that could generate PCR products related to transgene presence, rather than expression. Expression of angiostatin, endostatin and IFN-␣1 was detected exclusively in tumors generated by cells transduced with the corresponding retroviral vector, as shown in Figure 6. This experiment indicated that the transgenes were transcriptionally active in vivo, suggesting that the viral promoter was not silenced in this system. Although angiostatin did not delay tumor growth, it was expressed in both the MDAangio and the MCF-angio tumors, as confirmed by both RT-PCR analysis of tumor samples (Figure 6) and immunohistochemistry for the HA tag attached to the angiostatin fusion protein (Figure 7a–d) which showed strong, diffuse positivity in the mAST-transduced tumors. To rule out that angiostatin may undergo cleavage in vivo with loss of function, we analyzed tumor lysates by Western blotting with the anti-HA-specific antibody. As Gene Therapy

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

872

Figure 5 Histological and immunoistochemical analysis of the growing tumors 14 days after injection of 5 × 105 MDA-MB435 transduced cells. Panels (a–d), H&E staining, original magnification: ×20. Only the tumor sample obtained from IFN-␣1-producing cells (d) showed a large area of ischemic coagulative necrosis, while control MDA-LE cells (a), MDA-endo (b), and MDA-angio (c) injected mice show tumor masses composed of viable cells. Panels (e–h), immunohistochemical analysis with anti-CD31 mAb showed a reduced microvessel density in IFN-producing tumors (h) compared with control MDA-LE (e), MDA-angio (f), and MDA-endo (g) tumors.

shown in Figure 7e, full-length angiostatin was detected in tumors formed by MDA-MB435 and MCF7 cells. To investigate whether interferon was locally produced, we recovered the tumor cells 3 weeks after implantation, cultured them in vitro, and assayed cytokine levels in the culture medium. IFN-␣1 levels of 256– 512 U/ml were detected in these supernatants, indicating that the tumor cells continued to produce IFN-␣1 for at least 3 weeks after implantation. To confirm IFN proGene Therapy

duction in vivo, murine splenocytes were analyzed for the expression of the Ly-6c marker, which is up-modulated by this cytokine. These data (not shown) indicated that IFN-␣1 was produced in vivo by the tumor cells and released systemically. As assays to detect endostatin in serum are available, we investigated systemic endostatin expression. In animals which had received endostatin-producing cells serum endostatin levels increased with increasing tumor

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

Figure 6 Angiogenesis inhibitor expression within the tumors. RT-PCR analysis with transgene-specific primers confirmed angiostatin, endostatin and IFN-␣1 expression in the tumors formed by cells transduced with the respective vectors.

size, and 8 weeks after MDA-MB435 and MCF7 cell implantation were on average 287 ng/ml and 670 ng/ml, respectively (Figure 8). Interestingly, an increase was also observed in animals that had received control cells. This might represent endogenous production of endostatin in tumor-bearing animals (Figure 8). Indeed, basal levels in the different animal groups were similar, and about 30– 70 ng/ml as observed by others,15 and did not significantly increase over time, unless the animals were injected with tumor cells (data not shown).

Discussion Attention for angiostatic genes has recently been heightened by the possibility that they may be delivered by gene transfer methods. The goal of angiostatic gene transfer is to utilize some host cells as the ‘cell factories’ of the therapeutic molecule, thus circumventing the critical problems associated with long-term administration of protein-based drugs. Angiogenesis seems to represent an important prognostic factor in breast cancer,16,17 suggesting that its inhibition could play a role in controlling this disease, and leading us to test model systems of breast cancer. Previous reports indicated that selected angiostatic molecules delivered by gene transfer methods act as effective inhibitors of mammary tumor growth in mice.18–20 However, a direct comparison of different inhibitors was lacking. Our objective was to determine the sensitivity of two different mammary tumor systems to three endogenous inhibitors of angiogenesis delivered by viral vectors. This should help to define which agent might be more effective in this setting. Our findings indicate that mammary tumor cell lines implanted s.c. in nude mice were most sensitive to IFN␣1. Endostatin significantly retarded the growth of MDAMB435-derived tumors, but had little effect on MCF7derived tumors, while angiostatin did not exert any effect on tumors derived from either cell type. Interestingly, Kuo et al12 also observed heterogeneous inhibition of tumor growth by anti-angiogenic genes. These investigators used recombinant adenoviruses carrying the angiostatin, endostatin and soluble VEGF receptor genes to increase systemic levels of the angiostatic molecules in different tumor models. Evaluation of the in vitro biological activity of the angiogenesis inhibitors was based on endothelial cell proliferation and migration. Interestingly, only IFN-␣1 reduced

both HUVEC proliferation and migration in vitro, and showed maximal anti-tumor activity in vivo. Despite a lack of efficacy on breast tumors, angiostatin produced by the retroviral vector was biologically active in vitro, even though compared with IFN its effects on HUVEC proliferation and migration were less pronounced. The angiostatin used was fused to the HA epitope. This was necessary to detect the protein, as specific anti-murine angiostatin antibodies are not available. Although a comparison with native murine angiostatin is lacking, this HA-tagged murine angiostatin inhibited capillary endothelial cell proliferation and suppressed the growth of T241 fibrosarcoma cells.14 We recently showed that angiostatin delivery by the same vector significantly delayed the growth of Kaposi’s sarcoma cells in vivo, transduced by the same vector.21 Furthermore, other groups have used HA-tagged angiostatin for gene therapy purposes, with evidence of anti-angiogenic effects.19 This points to the conclusion that the tag does not interfere with angiostatin activity. Since all the angiostatic molecules used in this study were active on endothelial cells in vitro, we suggest that the heterogeneous anti-tumor effects observed in vivo might be related to the levels of angiostatin and endostatin. As expression levels can vary greatly among different tumors following gene delivery, they should be considered as a potential limiting factor for such approaches. In fact, the levels of angiostatin produced by transduced Kaposi’s cells, whose growth was inhibited by angiostatin,21 were approximately 10-fold higher than those produced by breast cancer cells, likely due to the different activity of the Mo-MLV promoter/enhancer driving transgene expression in the two cell lines. Thus, it is possible that critical angiostatin levels must be achieved for significant anti-tumor effects to be observed. The similar growth rate of endostatin-producing MCF7 tumors compared with controls, and angiostatin’s lack of effect on the growth of both cell lines did not appear to be due to gene silencing in vivo, which has been occasionally observed in previous studies.22 Indeed, transgene expression and local full-length angiostatin production were demonstrated in the tumors. Furthermore, serum endostatin levels up to 1 ␮g/ml were observed in animals that received endostatin-transduced MCF7 cells. Interestingly, endostatin showed divergent effects on the two breast cancer cell lines studied. These two lines differ in many genetic and phenotypic aspects. We observed that parental MCF7 cells grow much faster than MDA-MB435 cells in nude mice. Rapidly growing tumors may generate ischemic areas within the tumor mass at early time points, leading to up-regulation of pro-angiogenic ischemia-induced genes, such as VEGF that could favor tumor growth. Endostatin therapy, which in our study delayed the growth of MDA-MB435 but not MCF7 cells, might be particularly effective in slowly growing tumors. In keeping with these findings, a recent study of the effects of angiogenesis inhibitors on multistage pancreas carcinogenesis in mice showed that AGM-1470, angiostatin, BB-94 and endostatin had distinct efficacy profiles depending on the stage of tumor development and probably the kinetics of cell growth.23 In future studies, it will also be interesting to evaluate the possible correlation between status of estrogen receptors on breast cancer cells and response to endostatin. Our findings also indicated that IFN-␣1 is the most

873

Gene Therapy

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

874

Figure 7 Angiostatin expression in MCF7 and MDA-MB435 tumors. (a–d) Immunohistochemical analysis of tumor sections with anti-HA staining (mAb anti-HA, hematoxylin counterstain, original magnification ×20). HA immunostaining in one MCF7-angio tumor (a) and in one MDA-angio tumor (c). No specific staining was detected following incubation of the samples only with the secondary antibody (b and d). (e) Expression of angiostatin in MDA-MB435 and MCF7 tumors by Western blotting. Lysates from tumors formed by angiostatin- (mAST) and EGFP- (LE) transduced cells were analyzed: mAST-transduced tumors contained full-length angiostatin (58 kDa).

active single agent against the two breast cancer cell lines studied. IFN-␣ is used for the treatment of more than 14 types of cancer, including some hematological malignancies and certain solid tumors, such as melanoma, renal carcinoma and Kaposi’s sarcoma.7,24 Its anti-tumor activity can be attributed, at least in part, to its direct effects on tumor cells. In fact, in some cases IFN-␣ can directly inhibit the proliferation of normal and tumor cells in vitro and in vivo by modulating the expression of oncogenes or tumor suppressor genes, and it can increase MHC class I expression, thus favoring immune recognition.24 Our findings rule out a direct anti-proliferative effect of IFN-␣1 on MCF7 and MDA-MB435 cells. However, the possibility exists that some other cell types, Gene Therapy

including macrophages and NK cells, might contribute to the observed anti-tumor effect in these mice. In fact, type I IFNs exert multiple regulatory activities on host immune cells that are particularly relevant to the overall antitumor response.25,26 We did not detect a more intense lymphoid infiltrate in IFN-␣1-producing tumors compared with controls or angiostatin- or endostatin-producing tumors. Furthermore, we tested the susceptibility of MCF7 and MDA-MB435 cells to NK-mediated lysis in vitro, and found that neither was efficiently killed by activated murine NK cells (data not shown). It might be possible that some host cells contribute differently to the anti-tumor effect of IFN-␣1, ie by producing soluble factors that reduce tumor growth, or by affecting endothelial

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

vector encoding murine IFN-␣1, a cytokine known to exhibit some species specificity.32 However, we also analyzed the expression of murine bFGF in vivo by RT-PCR, and found similar levels of expression in all the tumors examined (data not shown) So, within the limitations of our PCR assay, we conclude that the angiogenesis inhibition observed in this system was not associated with marked down-regulated expression of angiogenic molecules in vivo. Since IFNs might exert angiostatic effects at low, constant doses, which are readily obtained by current gene delivery systems as shown here, it is tempting to speculate that local IFN-␣ production by genetically modified cells, either tumor cells as in this study or host non-neoplastic cells, might represent an effective way to delay tumor growth by inhibition of angiogenesis. The inclusion of this gene in lentiviral vectors will also enable us to test this hypothesis on established mammary tumors.

875

Materials and methods

Figure 8 Serum endostatin levels in nude mice following s.c. injection of endostatin-transduced cells. Serum samples were analyzed at regular intervals for endostatin content by an ELISA assay (Cytimmune) after injection of 5 × 105 endostatin- or LE-transduced (control) cells in nude mice. Endostatin levels were higher in the groups of mice that received endostatin-producing cells as compared with controls. An increase in endostatin serum levels was detected with both MDA-endo (a) and MCF-endo (b) cells.

cell function, as recently reported by Yao et al27 for IL12-activated murine NK cells. The inhibition of tumor-induced angiogenesis, and the induction of extensive ischemic necrosis in the tumor area by type I IFNs are well documented phenomena in animal models.28,29 Moreover, IFN-␣ was the first of the known angiogenesis inhibitors to be evaluated clinically and to show its therapeutic potential in angiogenic diseases, such as childhood hemangioma and Kaposi’s sarcoma.30,31 Its anti-angiogenic activity has been well established by a number of in vitro studies indicating that it can block endothelial cell migration,8,32 and also downregulate angiogenic factor expression, such as bFGF, IL-8, MMP-2 and MMP-9.33–36 IFN-␣ also down-regulates bFGF expression in human carcinomas.34 However, we did not observe reduced expression of bFGF mRNA in the IFN-␣1-producing MDA-MB435 cell line, neither in vitro (Figure 2c) nor in vivo (data not shown), as evaluated by RT-PCR experiments. This finding might be related to the fact that we transduced human breast cancer cells with a retroviral

Molecular cloning of angiostatin, endostatin, and IFN␣1encoding retroviral vectors The pMFG-mAST retroviral vector was generated by cloning a 1.5 kb-long NcoI/BamHI restriction fragment obtained from the digestion of pCMV-mAST,14 which encodes murine angiostatin with an HA tag attached as a fusion protein to the C-terminus, in the corresponding cloning sites of the MFG retroviral vector (Figure 1).37 Similarly, murine endostatin cDNA fused to the human B29 leader sequence was released from pB29ENDO,38 and cloned in the NcoI/BamHI sites of pMFG. These constructs were verified by PCR, restriction analysis, and sequencing before their further use. The LE retroviral vector used as control is a derivative of the LXSN retroviral vector,39 that carries an EGFP gene driven by the MoMLV LTR.40 The murine IFN-␣1-expressing retroviral vector was produced by a transfected GP+env Am12 packaging cell line, as previously described.13 Cell lines and transfections 293T human kidney cells, MCF7 and MDA-MB435 human breast cancer-derived cell lines were obtained from ATCC. All cell lines were grown in DMEM supplemented with 10% fetal calf serum (FCS) and 1% Lglutamine. Infectious particles carrying the angiostatin, the endostatin, and the control LE retroviral genomes were generated by a transient three-plasmid vector-packaging system, as previously described.41 Briefly, 293T cells were seeded in 100-mm-diameter Petri dishes at 5 × 106cells/dish, and transfected overnight according to a calcium-phosphate protocol using 12 ␮g of a plasmid DNA encoding the different retroviral vector genomes, 6 ␮g of a Mo-MLV gagpol expression construct, and 0.3 ␮g of a VSV-G expression plasmid, which encodes the envelope used by the vectors in this study. Thirty-six hours after transfection, culture medium was replaced with fresh DMEM without FCS. Twenty-four hours later the retroviral vector-containing supernatants were harvested, passed through 0.45-␮m filters, and stocked at –80°C until further use. Transduction of cells with retroviral vectors To assess the ability of the virions to transduce breast cancer cells, 1 ml of filtered supernatant was layered over Gene Therapy

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

876

MCF7 and MDA-MB435 target cells that had been seeded into six-well culture plates at 2 × 105 cells per well the day before infection. Protamine sulphate (8 ␮g/ml) (Sigma, St Louis, MO, USA) was added to the wells, and the cells were kept in a total volume of 2 ml. After 6–9 h at 37°C, 3 ml of medium were added to dilute the protamine sulphate; 36 h later, the cells were split 1:4 in 6-cm-diameter Petri dishes, and kept in culture until they underwent a new transduction cycle 24 h later. This procedure was repeated five times. The percentage of EGFP+ cells after each transduction cycle with the EGFP-encoding vector was determined by FACS analysis, and used as a parameter to estimate the genetically modified cell fraction which, after the fifth transduction cycle, was >90%. Western blotting Angiostatin and endostatin expression in supernatants was determined by Western blotting, according to standard protocols. For immunoblotting analysis, 40 ␮l FCSfree supernatant was electrophoresed on 10% polyacrylamide gels, and separated proteins were blotted for 2 h at 400 mA on to a nitrocellulose membrane. In a set of experiments with endostatin, the supernatant was concentrated 10-fold on Ultrafree columns (Millipore, Bedford, MA, USA) with 10 kDa cutoff. The membrane was then saturated with PBS 1% non-fat dry milk (Sigma) as blocking buffer for 3 h at RT, and then stored at –20°C until use. Immunoprobing was performed with a 1:8000 dilution of a murine mAb against the hemagglutinin (HA) tag (BabCo, Richmond, CA, USA), or a 1:5000 dilution of a rabbit antiserum against murine endostatin (Neosystem, Meylan, France), followed by hybridization with a 1:5000 diluted anti-mouse or anti-rabbit HRP-conjugated antibody (Amersham-Pharmacia, Little Chalfont, UK). Antigens were identified by luminescent visualization using the SuperSignal kit (Pierce, Rockford, IL, USA). Angiostatin and endostatin measurements Angiostatin levels in the supernatants of the transduced breast cancer cell lines were estimated by Western blotting and compared with an HA-tagged ␤-gal protein which was used as an internal standard. Endostatin levels in conditioned supernatants and serum were measured with a commercially available ELISA kit (Cytimmune Sciences, College Park, MD, USA). Cell viability assays We analyzed the viability of the genetically modified cells before in vivo experiments by conventional eosin staining, and a two-color fluorescence cell viability assay (LIVE/DEAD Kit, Molecular Probes, Leiden, The Netherlands), that allows quantitative analysis of live and dead cells by flow cytometry. To this end, 1 × 106 MDAMB435 and MCF7 cells were trypsinized, washed in PBS and stained for 10 min at room temperature in a 200 ␮l volume of 1 ␮M calcein AM and 8 ␮M ethidium homodimer-1. Following incubation, cell samples were analyzed by flow cytometry with excitation at 488 nm. The fluorescence emission was acquired separately at 530 nm and 585 nm for the live-cell and the dead-cell population, respectively. In vitro migration assays Chemotaxis assays were performed in Boyden chambers as previously described,42 using pvp-free polycarbonate

Gene Therapy

filters with 12 ␮m pores for HUVEC cells coated with 5 ␮g/ml of gelatin. 5 × 104 HUVECs in serum-free medium (SFM) containing 0.1% bovine serum albumin were placed in the upper compartment; the lower compartment was filled with either SFM as a negative control, or conditioned medium from NIH-3T3 cells as chemoattractant. To test the inhibitory activity of recombinant proteins, HUVEC were preincubated for 30 min with different amounts of DMEM, conditioned medium from MCF7 and MDA-MB435 cells transduced with the endostatin, angiostatin, IFN-␣1, or the control retroviral vectors. The chambers were then incubated at 37°C in 5% CO2 for 6 h. Cells remaining on the upper surface of the filter were then mechanically removed, and five to 10 random fields of cells which had migrated to the lower surface of each filter were counted after staining. Assays were performed in triplicate, and repeated at least three times. In vitro proliferation assays The effects of angiostatin, endostatin and IFN-␣1 on the in vitro proliferation of breast cancer and endothelial cells were tested using [3H]thymidine proliferation assays. In a set of experiments, MCF7 and MDA-MB435 cells, transduced with the different vectors, were plated in 96-multiwell plates at 1 × 104 and 5 × 103 cells/well, respectively. Cell proliferation was assessed at 24 h intervals over 5 consecutive days. At the appropriate time points, 1 ␮Ci/well of [3H]thymidine was added to the cultures and incorporation was quantified 24 h later. In experiments with endothelial cells, HUVEC were plated in collagen Icoated 96-multiwell plates at 800 cells/well. After overnight incubation at 37°C, the medium was aspirated and replaced with equal volumes (50 ␮l) of the conditioned medium from MCF7 and MDA-MB435 cells transduced with the different vectors. Eight replicates of each supernatant were tested. After 30 min incubation at 37°C, 100 ␮l of M199 medium containing 10% FCS and 10 ng/ml bFGF were added. After 72 h incubation at 37°C, HUVEC proliferation was analyzed by [3H]thymidine labelling (1 ␮Ci/well) and incorporation was quantified 24 h later. Tumor growth in vivo 5 x 105 retroviral vector-transduced MCF7 and MDAMB435 cells in Matrigel (10 mg/ml) were injected s.c. in the flanks of female nude (nu/nu) mice (Charles River Breeding Laboratories Italia, Calco, Italy). Tumor size was measured regularly over time with skin calipers. When the tumors reached pre-set dimensions according to current ethical practice, or 3 months after the beginning of the experiment, the animals were killed and tumors were collected, fixed in formalin, paraffin-embedded, sectioned, and stained with H&E. Procedures involving animals and their care conformed with institutional guidelines that comply with national and international laws and policies (EEC Council Directive 86/609, OJ L 358, 1, December 12, 1987; NIH Guide for the Care and Use of Laboratory Animals, NIH Publication 85-23, 1985). RT-PCR Total RNA was isolated from cell lines and tumors using the TRIzol reagent (Gibco BRL, Gaithersburg, MD, USA) according to the manufacturer’s instructions. One microgram of total RNA was digested with RNase-free

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

DNase I (Roche, Milan, Italy) and used for the synthesis of first-strand cDNA using oligo-dT primers and MLV reverse transcriptase (Applied Biosystems, Foster City, CA, USA) following the manufacturer’s protocol. DNase treatment was necessary to remove any residual genomic DNA from the RNA sample, which would have generated spurious bands at subsequent PCR analysis. Amplification was performed using the following primers: 5’-GGAACCCAGATGGAGAAACT-3’ (mAST-for) 5’-GGCTAGCGTAATCCGGAACAT-3’ (mAST-HA-rev) 5’-TGAAGATCTGTGACCATGGCCAGGCTGGCGTTGT-3’ (endo-for) 5’-CTATTTGGAGAAAGAGGTCATGAA-3’ (endo-rev) 5’-GGCTCTGTGCTTTCCTGATGG-3’ (IFN-for) 5’-CTCTTCTCTCAGTCTTCCCAGC-3’ (IFN-rev) 5’-CGAAGTGGTGAAGTTCATGGATG-3’ (VEGF-for) 5’-TTCTGTATCAGTCTTTCCTGGTGAG-3’ (VEGF-rev) 5’-ATGGCAGCCGGGAGCATCACC-3’ (bFGF-for) 5’-CACACACTCCTTTGATAGACACAA-3’ (bFGF-rev) 5’-ACCATTGGCAATGAGCGGTT-3’ (act-for) 5’-TCCTGCTTGCTGATCCACAT-3’ (act-rev)

PCR was performed for 35 cycles at 95°C 30 s, 56°C 30 s, 72°C 1 min. The amplified products were resolved by 1.5% agarose gel electrophoresis, and stained with ethidium bromide. No bands were obtained following PCR analysis of DNase-treated RNA samples amplified without RT. IFN titration IFN was titrated on murine L929 cells as described previously.43 Briefly, L929 cells were seeded at 2 × 104 cells/100 ␮l/well in 96-well plates in DMEM 2% FCS. One day later, 100 ␮l of test supernatant, or of a standard mouse IFN-␣/␤ preparation were added in duplicate, and serially diluted two-fold. One day later, the medium was removed from the wells and 100 ␮l of a vesicular stomatitis virus suspension (multiplicity of infection = 0.05 p.f.u./cell) were added to the cells. After 1 h incubation at 37°C in 5% CO2 atmosphere, the medium was removed and 100 ␮l of fresh DMEM 2% FCS were added. The cytopathic effect was observed under a light microscope 24 and 48 h later. One unit of IFN was defined as the amount necessary to inhibit 50% of the cytopathic effect. IFN titers are expressed as IU. Expression of the Ly-6C antigen The membrane expression of the Ly-6C antigen on mouse splenocytes was determined by flow cytometric analysis following labeling with a rat anti-murine Ly6C mAb, as described elsewhere.44 Fluorescein-isothiocyanate-conjugated antibody against rat Ig was purchased from KpL (Gaithersburg, MD, USA). Histology and immunohistochemistry Four-␮m thick sections were rehydrated, and either stained with H&E or processed for immunohistochemistry by standard procedures. For immunohistochemistry, rehydrated sections were blocked with irrelevant serum, followed by incubation with primary antibody: immunostaining was performed using the avidin-biotinperoxidase complex technique and 3-3’ diaminobenzidine as chromogen (Vector Laboratories, Burlingame, CA, USA). Sections were then lightly counter-stained with Mayer’s hematoxylin. The primary antibodies used were

a monoclonal anti-CD31 (Becton-Dickinson, Franklin Lakes, NJ, USA) to highlight vessels, or a monoclonal anti-HA tag (BabCo) to localize the HA-angiostatin fusion protein. Parallel negative controls without primary antibodies were run in all cases. Microvessel areas were quantified by manual counting of hotspots in sections. Ten representative fields for each tumor were counted.

877

Statistical analysis The significance of the difference in tumor size was determined by Student’s t test.

Acknowledgements We thank Dr E Lechman for the MFG retroviral vector; Dr E Shewach for the anti-Ly-6C antibody; Dr V Tosello for cytofluorimetric analysis; Mr P Gallo for artwork and Ms P Segato for help in the preparation of the manuscript. This work was supported by grants from the MURST 40% and 60%, the Italian Association for Cancer Research (AIRC), the Ministero della Sanita` Programma Nazionale Ricerca sull’AIDS, the Italian Foundation for Cancer Research (FIRC), the Fondazione Cassa di Risparmio di Padova e Rovigo, the CNR PF Biotecnologie.

References 1 Kong HL, Crystal RG. Gene therapy strategies for tumor antiangiogenesis. J Natl Cancer Inst 1998; 90: 273–286. 2 Cao Y. Endogenous angiogenesis inhibitors and their therapeutic implications. Int J Biochem Cell Biol 2001; 33: 357–369. 3 O’Reilly MS et al. Angiostatin: a novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma. Cell 1994; 79: 315–328. 4 O’Reilly MS et al. Endostatin: an endogenous inhibitor of angiogenesis and tumor growth. Cell 1997; 88: 277–285. 5 O’Reilly MS, Holmgren L, Chen C, Folkman J. Angiostatin induces and sustains dormancy of human primary tumors in mice. Nat Med 1996; 2: 689–692. 6 Boehm T, Folkman J, Browder T, O’Reilly MS. Antiangiogenic therapy of experimental cancer does not induce acquired drug resistance. Nature 1997; 390: 404–407. 7 Gutterman JU. Cytokine therapeutics: lessons from interferon alpha. Proc Natl Acad Sci USA 1994; 91: 1198–1205. 8 Brouty-Boye D, Zetter BR. Inhibition of cell motility by interferon. Science 1980; 208: 516–518. 9 Slaton JW et al. Interferon-alpha-mediated down-regulation of angiogenesis-related genes and therapy of bladder cancer are dependent on optimization of biological dose and schedule. Clin Cancer Res 1999; 5: 2726–2734. 10 Ferrantini M, Belardelli F. Gene therapy of cancer with interferon: lessons from tumor models and perspectives for clinical applications. Semin Cancer Biol 2000; 10: 145–157. 11 Albini A et al. Inhibition of angiogenesis and vascular tumor growth by interferon-producing cells: a gene therapy approach. Am J Pathol 2000; 156: 1381–1393. 12 Kuo CJ et al. Comparative evaluation of the antitumor activity of antiangiogenic proteins delivered by gene transfer. Proc Natl Acad Sci USA 2001; 98: 4605–4610. 13 Rozera C et al. Interferon (IFN)-beta gene transfer into TS/A adenocarcinoma cells and comparison with IFN-alpha: differential effects on tumorigenicity and host response. Am J Pathol 1999; 154: 1211–1222. 14 Cao Y et al. Expression of angiostatin cDNA in a murine fibrosarcoma suppresses primary tumor growth and produces longterm dormancy of metastases. J Clin Invest 1998; 101: 1055–1063. 15 Feldman AL et al. Antiangiogenic gene therapy of cancer utilizing a recombinant adenovirus to elevate systemic endostatin levels in mice. Cancer Res 2000; 60: 1503–1506. Gene Therapy

Angiogenesis inhibitors gene transfer in breast cancer cells S Indraccolo et al

878

16 Gasparini G, Harris AL. Clinical importance of the determination of tumor angiogenesis in breast carcinoma: much more than a new prognostic tool. J Clin Oncol 1995; 13: 765–782. 17 Gasparini G et al. Prognostic significance of vascular endothelial growth factor protein in node-negative breast carcinoma. J Natl Cancer Inst 1997; 89: 139–147. 18 Chen QR, Kumar D, Stass SA, Mixson AJ. Liposomes complexed to plasmids encoding angiostatin and endostatin inhibit breast cancer in nude mice. Cancer Res 1999; 59: 3308–3312. 19 Sacco MG et al. Liposome-delivered angiostatin strongly inhibits tumor growth and metastasization in a transgenic model of spontaneous breast cancer. Cancer Res 2000; 60: 2660–2665. 20 Sauter BV et al. Adenovirus-mediated gene transfer of endostatin in vivo results in high level of transgene expression and inhibition of tumor growth and metastases. Proc Natl Acad Sci USA 2000; 97: 4802–4807. 21 Indraccolo S et al. Effects of angiostatin gene transfer on functional properties and in vivo growth of Kaposi’s sarcoma cells. Cancer Res 2001; 61: 5441–5446. 22 Palmer TD, Rosman GJ, Osborne WR, Miller AD. Genetically modified skin fibroblasts persist long after transplantation but gradually inactivate introduced genes. Proc Natl Acad Sci USA 1991; 88: 1330–1334. 23 Bergers G et al. Effects of angiogenesis inhibitors on multistage carcinogenesis in mice. Science 1999; 284: 808–812. 24 Pfeffer LM et al. Biological properties of recombinant alphainterferons: 40th anniversary of the discovery of interferons. Cancer Res 1998; 58: 2489–2499. 25 Belardelli F. Role of interferons and other cytokines in the regulation of the immune response. APMIS 1995; 103: 161–179. 26 Belardelli F, Gresser I. The neglected role of type I interferon in the T-cell response: implications for its clinical use. Immunol Today 1996; 17: 369–372. 27 Yao L et al. Contribution of natural killer cells to inhibition of angiogenesis by interleukin-12. Blood 1999; 93: 1612–1621. 28 Dvorak HF, Gresser I. Microvascular injury in pathogenesis of interferon-induced necrosis of subcutaneous tumors in mice. J Natl Cancer Inst 1989; 81: 497–502. 29 Thomas H, Balkwill FR. Effects of interferons and other cytokines on tumors in animals: a review. Pharmacol Ther 1991; 52: 307–330. 30 Ezekowitz RA, Mulliken JB, Folkman J. Interferon alfa-2a therapy for life-threatening hemangiomas of infancy. N Engl J Med 1992; 326: 1456–1463.

Gene Therapy

31 Krown SE. Interferon-alpha: evolving therapy for AIDS-associated Kaposi’s sarcoma. J Interfer Cytok Res 1998; 18: 209–214. 32 Sidky YA, Borden EC. Inhibition of angiogenesis by interferons: effects on tumor- and lymphocyte-induced vascular responses. Cancer Res 1987; 47: 5155–5161. 33 Dinney CP et al. Inhibition of basic fibroblast growth factor expression, angiogenesis, and growth of human bladder carcinoma in mice by systemic interferon-alpha administration. Cancer Res 1998; 58: 808–814. 34 Singh RK et al. Interferons alpha and beta down-regulate the expression of basic fibroblast growth factor in human carcinomas. Proc Natl Acad Sci USA 1995; 92: 4562–4566. 35 Oliveira IC, Sciavolino PJ, Lee TH, Vilcek J. Downregulation of interleukin 8 gene expression in human fibroblasts: unique mechanism of transcriptional inhibition by interferon. Proc Natl Acad Sci USA 1992; 89: 9049–9053. 36 Gohji K et al. Human recombinant interferons-beta and -gamma decrease gelatinase production and invasion by human KG-2 renal-carcinoma cells. Int J Cancer 1994; 58: 380–384. 37 Krall W, Kohn DB. Expression levels by retroviral vectors based upon the N2 and the MFG backbones. Gene Therapy 1996; 3: 365. 38 Indraccolo S et al. Generation of expression plasmids for angiostatin, endostatin and TIMP-2 for cancer gene therapy. Int J Biol Markers 1999; 14: 251–256. 39 Miller AD, Rosman GJ. Improved retroviral vectors for gene transfer and expression. Biotechniques 1989; 7: 980–982. 40 Klein D et al. Rapid identification of viable retrovirus-transduced cells using the green fluorescent protein as a marker. Gene Therapy 1997; 4: 1256–1260. 41 Indraccolo S et al. Pseudotyping of Moloney leukemia virusbased retroviral vectors with simian immunodeficiency virus envelope leads to targeted infection of human CD4+ lymphoid cells. Gene Therapy 1998; 5: 209–217. 42 Albini A et al. A rapid in vitro assay for quantitating the invasive potential of tumor cells. Cancer Res 1987; 47: 3239–3245. 43 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. 44 De Giovanni C et al. Immunological and non-immunological influence of H-2Kb gene transfection on the metastatic ability of B16 melanoma cells. Int J Cancer 1991; 48: 270–276.