Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Granulocyte-macrophage colony-stimulating factor (GM-CSF) secreted by cDNA-transfected tumor cells induces a more potent antitumor response than exogenous GM-CSF Fu-Shun Shi,1 Sharon Weber,1 Jacek Gan,2 Alexander L. Rakhmilevich,2 and David M. Mahvi1 Departments of 1Surgery and 2Human Oncology, University of Wisconsin, Madison, Wisconsin 53792. Clinical cancer gene therapy trials have generally focused on the transfer of cytokine cDNA to tumor cells ex vivo and with the subsequent vaccination of the patient with these genetically altered tumor cells. This approach results in high local cytokine concentrations that may account for the efficacy of this technique in animal models. We hypothesized that the expression of certain cytokines by tumor cells would be a superior immune stimulant when compared with local delivery of exogenous cytokines. Granulocyte-macrophage colony-stimulating factor (GM-CSF) cDNA in a nonviral expression vector was inserted into MDA-MB231 (human breast cancer), M21 (human melanoma), B16 (murine melanoma), and P815 (mastocytoma) cells by particle-mediated gene transfer. The ability of transfected tumor cells to generate a tumor-specific immune response was evaluated in an in vitro mixed lymphocyte-tumor cell assay and in an in vivo murine tumor protection model. Peripheral blood lymphocytes cocultured with human GM-CSF-transfected tumor cells were 3- to 5-fold more effective at lysis of the parental tumor cells than were peripheral blood lymphocytes incubated with irradiated tumor cells and exogenous human GM-CSF. Mice immunized with murine GM-CSF-transfected irradiated B16 murine melanoma cells or P815 mastocytoma cells were protected from subsequent tumor challenge, whereas mice immunized with the nontransfected tumors and cutaneous transfection of murine GM-CSF cDNA at the vaccination site developed tumors more frequently. The results indicate that GM-CSF protein expressed in human and murine tumor cells is a superior antitumor immune stimulant compared with exogenous GM-CSF in the tumor microenvironment. Key words: Cytokines; granulocyte-macrophage colony-stimulating factor; gene transfer; tumor cells. T he development of cancer immunotherapy employing cytokines has been hampered by the inability to achieve tumoricidal cytokine levels without dose-limiting toxicity.1–3 However, systemic cytokine administration may not be necessary for the destruction of metastatic tumors. High peritumoral levels of cytokines produced from genes introduced into tumor cells can attract immune cells such as T lymphocytes, natural killer cells, neutrophils, and macrophages to the tumor site without systemic toxicity.4,5 The putative mechanism of cytokinemediated tumor destruction is not understood completely but often depends upon the generation of stimulated immune effector cells capable of tumor lysis.6 – 8 A number of animal studies have confirmed that vaccination with tumor cells transfected with cytokine genes can protect animals from subsequent challenge with wildtype tumors.9 –18 Thus, we19 and others20 –22 have shown Received December 12, 1997; accepted May 4, 1998. Address correspondence and reprint requests to Dr. David M. Mahvi, University of Wisconsin Hospitals and Clinics, H4/726 CSC, 600 Highland Avenue, Madison, WI 53792. © 1999 Stockton Press 0929-1903/99/$12.00/10 Cancer Gene Therapy, Vol 6, No 1, 1999: pp 81– 88 that the approach of vaccinating mice with autologous tumor cells that express transgenic granulocyte-macrophage colony-stimulating factor (GM-CSF) results in protection from subsequent tumor challenge. Numerous clinical trials have been initiated recently using cytokine gene-modified autologous tumor cells or tumor-infiltrating lymphocytes for the treatment of advanced cancer patients.23–26 Data from these studies imply that high local cytokine levels in the vicinity of tumor cells are as effective as systemic cytokine administration; the concentration of the cytokine in the tumor microenvironment is the critical factor.27 Previous clinical gene therapy protocols have relied on the delivery of cytokine DNA into tumor samples, with subsequent placement of these altered tumor cells as an antitumor vaccine.24 –26 This technique requires the insertion of cytokine genes in appropriate expression vectors into autologous tumors and has some limitations.28 An alternate method that would supply a high local cytokine concentration in the vicinity of autologous tumors is cytokine gene delivery to nontumor cells in the vicinity of tumor cells, such as particle-mediated gene transfer to the skin over a tumor vaccine with the gene gun. The gene gun can directly deliver ;10,000 DNA copies 81 82 SHI, WEBER, GAN, ET AL: GM-CSF SECRETED BY cDNA TUMOR CELLS per cell into the skin with minimal tissue damage and thus is ideal to test this alternate method. The therapeutic effect of this approach against established intradermal (i.d.) tumors has been demonstrated recently.29 We hypothesized that cytokine expression by the tumor cell would be a more effective immune stimulant than the recombinant cytokine protein delivered exogenously to the microenvironment of the tumor. We chose to investigate this effect in a human in vitro mixed lymphocyte-tumor cell culture system and in vivo in a murine tumor protection model. skin transfections. The DNA/gold particle preparation was coated and distributed evenly onto the inner surface of Tefzel tubing using a semiautomatic tube loader (Geniva); the tubing was cut into 0.5-inch lengths to ensure the delivery of 0.5 mg of gold and 1.25 mg of plasmid DNA of each cytokine per transfection. For the transfection of tumor cell lines, cells were g-irradiated at 10,000 rad using a 137Cs source. A total of 1.0 3 106 tumor cells in 20 mL of medium were spread into a 1.5-cm diameter smear in a 35-mm dish and immediately transfected with a 240-lb per square inch helium gas pulse followed by the addition of 2 mL culture medium. MATERIALS AND METHODS After human tumor cells (MDA and M21) were transfected, the medium was completely replenished with 2 mL of fresh medium in duplicate every day for 6 days. Cell culture supernatant was collected at the indicated timepoints and frozen at 220°C until assayed. B16 cells were transfected with mGMCSF coated beads, and supernatants were collected after a 24-hour incubation. Transgenic hGM-CSF assays were performed using an hGM-CSF kit (Biosource, Camarillo, Calif) according to the manufacturer’s instructions. The sensitivity of the assay was 15.6 pg/mL. Additional supernatants of cells transfected with hGM-CSF gene were also collected and frozen at 220°C for later bioassay. For mGM-CSF, a similar ELISA kit was prepared using commercial reagents (PharMingen, San Diego, Calif). The sensitivity of the assay was 156 pg/mL. All samples were assayed in duplicate. Tumor cell lines The MDA-MB-231 human breast cancer cell line (purchased from American Type Culture Collection, Rockville, Md), M21 human melanoma cell line (obtained from Dr. R. Reisfeld, The Scripps Research Institute, La Jolla, Calif), B16 mouse melanoma cell line,19 and P815 mouse mastocytoma cell line29 were used in this study. All cell lines were grown and maintained in RPMI 1640 cell culture medium (BioWhittaker, Walkersville, Md) supplemented with 10% heated-inactivated fetal bovine serum (Sigma, St. Louis, Mo), 2 mM L-glutamine, 1 mM sodium pyruvate, 1% nonessential amino acids (all from BioWhittaker), and 10 mg/mL ciprofloxacin (Miles, Kankakee, Ill). The RPMI 1640 culture medium with supplements (referred to as complete RPMI 1640 medium) was also used for culturing peripheral blood lymphocytes (PBLs). TF-1 human erythroleukemia cell line was maintained in complete RPMI 1640 medium containing 5 ng/mL human GM-CSF (hGMCSF) (obtained from Immunex, Seattle, Wash). Plasmids The hGM-CSF coding region including the coding sequences to nucleotide residues 27 to 1433 were amplified via polymerase chain reaction using wild-type hGM-CSF (a gift of Dr. James S. Malter, University of Wisconsin) as a template. To create the hGM-CSF expression plasmid pWRG3218, amplified cDNA was subcloned into a pUC19 vector containing a cytomegalovirus (CMV) immediate/early promoter/enhancer, a simian virus 40 intron sequence, and a bovine growth hormone polyadenylation region. The murine GM-CSF (mGM-CSF) expression plasmid pWRG3226 was created by replacing the lacZ gene in pCMV-b (Clontech, Palo Alto, Calif) with the mGM-CSF coding region from pGM3.2 (a gift of Dr. Nicholas Gough, Walter and Elisa Hall Institute, Melbourne, Australia). Both hGM-CSF and mGM-CSF plasmids contain the kanamycin-resistance gene. The luciferase (Lux) vector pCMV-luc30 was used as the control DNA. DNA transfections A helium gas-driven gene gun designed by D. McCabe (Geniva, presently PowderJect Vaccines, Madison, Wis) was used for DNA transfections. The plasmid DNA of cytokines was purified on Qiagen columns (Qiagen, Chatsworth, Calif) and precipitated onto 2-mm diameter gold particles at a density of 2.5 mg of DNA/mg of gold particles using the spermidine/ CaCl2 coating method.31 Particles were resuspended in a solution of 0.015 mg of polyvinyl pyrrolidone per milliliter in absolute ethanol for in vitro and ex vivo transfections and in 0.1 mg of polyvinyl pyrrolidone per milliliter of ethanol for direct Measurement of transgenic cytokine levels by enzymelinked immunosorbent assay (ELISA) hGM-CSF bioassay To determine whether the hGM-CSF detected by ELISA in culture supernatants from transfected human cells was biologically active, a TF-1 bioassay was performed. TF-1 cells (provided by T. Kitamura, University of Tokyo, Tokyo, Japan) were derived from a patient with erythroleukemia and were depend upon hGM-CSF or interleukin-3 for sustained proliferation.32 TF-1 cells were harvested from culture flasks and washed two times with plain RPMI 1640 medium. Serial dilutions of 100 mL of recombinant hGM-CSF (rhGM-CSF) or 100 mL of sample supernatants from the cells transfected with hGM-CSF were mixed with 100 mL of TF-1 cell suspension (5 3 103 cells/well) in complete RPMI 1640 medium. After 48 hours, 1 mCi of [3H]thymidine was added to each well, and the cells were incubated for an additional 18 hours at 37°C and 5% CO2. Cells were harvested using a semiautomatic cell harvester (Micromate 196, Packard, Downers Grove, Ill), and radioactivity was counted (Matrix 96, Packard). In vitro activation of PBLs Peripheral blood was collected from healthy donors and diluted with Hanks’ balanced salt solution (BioWhittaker); PBLs were separated on a Lymphoprep (Nycomed Pharma AS, Oslo, Norway). A total of 2 3 106 PBLs in 1 mL of complete RPMI 1640 medium were mixed with 1 3 106 irradiated tumor cells (MDA-MB-231 or M21) transfected with the hGM-CSF gene or the Lux gene as described previously or nontransfected with or without the addition of rhGM-CSF protein (65 ng); the PBLs were subsequently cultured at 37°C with 5% CO2 for 6 days. Before performing the PBL activation experiment, we transfected 1 3 106 irradiated tumor cells (MDA-MB-231 or M21) and incubated the transfected tumor cells at 37°C and 5% CO2 for 24 hours. The supernatants of transfected tumor cells were collected to determine the amount of transgenic Cancer Gene Therapy, Vol 6, No 1, 1999 SHI, WEBER, GAN, ET AL: GM-CSF SECRETED BY cDNA TUMOR CELLS 83 GM-CSF production by ELISA. The amount of exogenous rhGM-CSF protein used to activate PBLs was based on this measurement of the amount of transgenic GM-CSF present in the supernatant after 24 hours of incubation per 106 transfected cells. Cell-mediated cytotoxicity of human lymphocytes After culture for 6 days, PBLs were collected from culture flasks and used as effector cells in 4-hour chromium release assays against MDA-MB-23133 or M21 targets.34 Briefly, target cells were labeled with 250 mCi of 51Cr for 2 hours at 37°C. A total of 105 effector cells in 100 mL of complete RPMI 1640 medium were diluted serially. A total of 5 3 103 labeled target cells were then added to test wells in triplicate to produce effector to target ratios of 20:1, 10:1, 5:1, and 2.5:1. Plates were centrifuged at 100 3 g for 3 minutes and then incubated for 4 hours at 37°C. Supernatants were harvested using the Skatron harvesting system (Skatron, Sterling, Va) and counted for 1 minute in a g counter. The percentage of cytotoxicity mediated by the PBLs was calculated as follows: % of cytotoxicity 5 ([exp. counts per minute (cpm) 2 spon. cpm]/[max. cpm 2 spon. cpm]) 3 100, where exp. was the experimental number of counts from target cells incubated with effectors, spon. was the spontaneously released counts obtained from targets cultured in medium alone, and max. was the maximum counts obtained from target cells lysed completely with a 2% centrimide solution (Sigma). 51Cr release data from all four effector to target ratios were also converted to lytic units (LU). In this study, 1 LU was defined as the number of harvested effector cells resulting in a 20% lysis of 5 3 103 target cells; LU are expressed as LU/107 effector cells harvested. Murine studies Female C57BL/6 and DBA/2 female mice (6 – 8 weeks of age) were obtained from the Harlan-Sprague-Dawley Animal Facility (Indianapolis, Ind) and Taconic Farms (Germantown, NY), respectively. All animal experiments were conducted in accordance with the principles stated in the Guide for the Care and Use of Laboratory Animals. (Publication 86-23, National Institutes of Health, Bethesda, Md, 1985). The tumors used in this study were B16 melanoma and P815 mastocytoma, syngeneic in C57BL/6 and DBA/2 mice, respectively. For tumor vaccination, mice were injected i.d. on day 0 (P815 tumor) or on days 0 and 7 (B16 tumor) in the right shaved abdominal area with 1 3 106 g-irradiated B16 melanoma cells or P815 mastocytoma tumor cells; B16 cells transfected with mGM-CSF were injected at a dose of 5 3 105. On each day, mice received either (a) an i.d. injection of mGMCSF-transfected B16 or P815 tumor cells, (b) nontransfected tumor cells administered as a vaccine with a cutaneous transfection of mGM-CSF cDNA to the skin over a vaccine site, (c) nontransfected cells with a cutaneous transfection with Lux cDNA, or (d) nontransfected cells without cutaneous transfections. In group 1, mice received a single injection of irradiated mGM-CSF-transfected tumor cells on each day, whereas in groups 2 and 3, cutaneous treatments were performed at four sites on the abdominal vaccine injection area on each treatment day. Mice were challenged in the left side of the abdomen with 105 B16 cells or 5 3 104 P815 cells at 2 weeks or 1 week after the last vaccination, respectively. Tumor diameter was measured two times per week. Cancer Gene Therapy, Vol 6, No 1, 1999 Figure 1. hGM-CSF production by transfected, irradiated MDA-MB231 (f) and M21 cells (F). After irradiation (10,000 rad), cells were transfected with GM-CSF expression plasmids using the gene gun. At 24-hour intervals, supernatants were collected for ELISA and the medium was exchanged with 2 mL of fresh complete RPMI 1640 medium. Therefore, only the production of transgenic GM-CSF from the previous day was measured. No cells were replaced or removed from the originally transfected dishes. The results are expressed in nanograms of cytokine/106 initially treated cells/24 hours. RESULTS Transfection of human tumor cell lines results in biologically active GM-CSF expression The transfection of MDA-MB-231 and M21 with hGMCSF DNA resulted in GM-CSF protein expression of 72.2 and 60.0 ng/106 cells after 24 hours of culture, respectively. Transfected cells demonstrated GM-CSF protein expression that persisted for 6 days (Fig 1). To confirm that the GM-CSF detected by ELISA was biologically active, the supernatants of transfected cells were added to a GM-CSF-dependent cell line, TF-1. These cells, which were grown in the presence of supernatant from GM-CSF-transfected cells, proliferated as measured by thymidine incorporation. In contrast, no proliferation of TF-1 cells was observed in the presence of supernatants from nontransfected cells or transfected cells with the Lux gene (Fig 2). The mean response of TF-1 to supernatants of hGM-CSF-transfected MDA and M21 cells was 6600 cpm and 4300 cpm, respectively, whereas the mean response to supernatants of untransfected or Lux-transfected MDA and M21 cells was 85 cpm and 6.1 cpm, respectively. PBLs stimulated in vitro with GM-CSF-transfected tumor cells killed parental tumor cells The cytotoxic activity of human PBLs was assessed after stimulation by hGM-CSF-transfected, Lux-transfected tumor cells or by nontransfected tumor cells plus rhGMCSF protein. PBLs exposed to either MDA-MB-231 or M21 tumor cells transgenic for hGM-CSF were superior effectors when compared with the addition of exogenous recombinant protein (Fig 3). Minimal lytic activity was observed when PBLs were incubated with irradiated tumor cells alone. The addition of GM-CSF protein to the culture supernatant was similar to the tumor cells alone without GM-CSF. The absolute lytic activity of the 84 SHI, WEBER, GAN, ET AL: GM-CSF SECRETED BY cDNA TUMOR CELLS Figure 2. Biological activity of hGM-CSF in the supernatants of hGM-CSF-transfected MDA-MB-231 and M21 cells. Supernatants were harvested after transfected cells had been incubated for 24 hours, and proliferation activity on TF-1 cells was analyzed using the ]3H[thymidine incorporation assay. The results indicated that supernatants from hGM-CSF-transfected cells contained biologically active hGM-CSF, whereas Lux-transfected cells contained no hGMCSF. The values reported are the mean cpm of triplicate wells. stimulated effector cells varied, but the PBLs stimulated with tumor cells transfected with GM-CSF cDNA were superior to control PBL populations in all samples tested. Transgenic GM-CSF was a more potent immunogen than exogenous GM-CSF protein in vivo The amount of GM-CSF protein in the local microenvironment was determined before the in vivo vaccination studies. To titrate the dose of GM-CSF protein, murine skin was evaluated 24 hours after either a single cutaneous treatment over a B16 tumor site or an injection of B16 transfected ex vivo with an identical GM-CSF cDNA expression vector (Fig 4). Tissue levels of mGM-CSF for nontransfected cells (group 1) and a cutaneous transfection with the Lux DNA (group 2) over the top of nontransfected tumor cells were ,0.1 ng of tissue per site. Cutaneous transfection with a single dose of mGMCSF over the top of irradiated but nontransfected tumors resulted in 1.5 ng of tissue per site (group 3). A total of 6.5 ng of tissue per site was present after i.d. injection of mGM-CSF-transfected B16 tumor vaccine alone (group 4). Because the expression of transgenic GM-CSF after cutaneous transfection (1.5 ng of tissue per site/24 hours) seemed to be approximately four times lower than after ex vivo tumor cell transfection, we performed cutaneous transfections for tumor vaccination experiments at four sites, as indicated in the legend to Figure 5B. The results of two tumor vaccination experiments using two different tumor models are presented in Figure 5. In the P815 tumor model (Fig 5A), mice were vaccinated once and challenged with the parental tumor 1 week later. The results show that delivery of the GM-CSF gene into the skin above the nontransfected tumor vaccination site resulted in protection in 16.6% of Figure 3. Lytic activity of healthy donor PBLs cultured without tumor cells (PBLs alone), cocultured with tumor cells as stimulator cells (PBLs 1 SC), cocultured with GM-CSF-transfected tumor cells (PBLs 1 SC-GM-CSF), or cocultured with tumor cells plus 65 ng of exogenous rhGM-CSF (PBLs 1 rhGM-CSF). Two different tumor cell lines (MDA-MB-231 (A) and M21 (B)) were used as both stimulators and targets with five (A) and six (B) distinct PBL populations. mice, whereas 33.3% of mice immunized with ex vivo GM-CSF-transfected tumor cells were protected from a subsequent tumor challenge. The protection of animals vaccinated with GM-CSF-transfected tumors was significantly (P , .022) greater than that seen in mice vaccinated with control-transfected tumors, whereas the delivery of GM-CSF cDNA to the skin over the tumor was not significantly different from the control-transfected cohort (Mantel-Haenszel x2 test differences in time until developing tumor). Comparison of ex vivo GM-CSF-transfected tumor vaccination with GM-CSF cDNA transfection of the skin over the tumor failed to reach statistical significance. To confirm and extend these results, we used a B16 melanoma model in the next Cancer Gene Therapy, Vol 6, No 1, 1999 SHI, WEBER, GAN, ET AL: GM-CSF SECRETED BY cDNA TUMOR CELLS 85 Figure 4. mGM-CSF levels detected in extracts from tissue biopsies of vaccination sites after C57BL/6 mice were injected with a transgenic B16 cell vaccine and/or subjected to cutaneous mGMCSF transfection. At 24 hours after the first vaccination or cutaneous transfection, two mice in each group were sacrificed for an evaluation of the in vivo tissue level of mGM-CSF. Data show the mean levels of mGM-CSF in a vaccination site (a standard 1.5 3 1.5 cm2 full thickness skin biopsy). experiment.19 To augment the antitumor effect, we also used two vaccinations instead of one; each vaccination was performed at four sites. The results in Figure 5B show that the nontransfected tumor vaccine alone or in combination with cutaneous transfection with GM-CSF cDNA did not result in protection from a B16 tumor challenge. In contrast, vaccination with ex vivo GM-CSFtransfected tumor cells resulted in the protection of 62.5% mice from an identical B16 tumor challenge. In this experimental system, ex vivo GM-CSF-transfected tumor vaccination was statistically superior to GM-CSF cDNA transfection of the skin over the tumor (P , .007). DISCUSSION Initial animal models for the therapy of malignancies by vaccination with tumor cells transfected with cytokine genes were based on the hypothesis that a high local cytokine concentration in the vicinity of autologous tumor cells was necessary for the protection of animals from subsequent tumor challenge. This approach was thought to be more effective than systemic cytokine administration because higher local cytokine levels in the vicinity of the tumor could be generated without the toxicity of systemic cytokine administration.35 The most direct method for cytokine delivery to the tumor microenvironment has been the in vitro transfection of tumor cells, with subsequent placement of this “tumor vaccine” i.d. or subcutaneously.4 –5,8,20 Consequently, the expression of the transgene by the tumor cells may be a significant factor in the presence of stimulation of an antitumor immune response. Many cytokine genes have been studied in animal tumor model systems and human cancer clinical trials.8,9,20,23,26,29,36 When different cytokine genes were compared, GM-CSF most effectively prevented subse- Cancer Gene Therapy, Vol 6, No 1, 1999 Figure 5. Effect of transgenic mGM-CSF production on protection from tumor challenge after vaccination. A: Mice were not vaccinated (F) or were injected i.d. with irradiated P815 tumor cells. The skin above the tumor cell implants was transfected with GM-CSF cDNA (‚) or left untreated (E). In another group, irradiated P815 tumor cells were transfected ex vivo with mGM-CSF cDNA and injected i.d. at a dose of 106 cells (F). At 1 week after vaccination, all mice were challenged i.d. with unmodified P815 cells and observed for tumor growth. Data are presented for six mice per group. B: Mice were not vaccinated (F) or were injected i.d. with irradiated B16 tumor cells on days 0 and 7. The skin above each tumor cell implant was transfected with GM-CSF cDNA (‚) or Lux cDNA (E). In another group, irradiated B16 tumor cells were transfected ex vivo with mGM-CSF cDNA right and injected on days 0 and 7 (F). At 2 weeks after the last vaccination, mice were challenged i.d. with parental B16 cells and observed for tumor growth. Data are presented for eight mice per group. quent tumor growth in a poorly immunogenic tumor model.20 We confirmed these findings in a similar protection model.19 Therefore, the present study focused on the most effective cytokine gene in this model system. We demonstrated that hGM-CSF expression by human tumor cell lines more effectively stimulated PBLs to kill the parental tumor. This effect was consistent in PBL populations tested. The length of expression by both tested tumor cells lasted $6 days, which led to a more prolonged exposure of effector cells to the cytokine. This 86 long-lasting although transient19 production of GM-CSF by irradiated tumor cells was consistent with data reported by Rosenthal et al,37 who showed that irradiated gene-transfected fibroblasts produced a detectable level of G-CSF for 12 days and had systemic hematological effects. mGM-CSF was detected at both the injection site of i.d. tumor vaccination and at the site of direct skin bombardment, which confirms that the particle-mediated gene transfer of nonviral expression vectors can deliver genetic material directly into cells both in vitro and in vivo.38 – 40 The latency of gene expression and thus protein expression may result in more effective stimulation of the naive PBL populations studied. An alternative explanation may be that higher local GM-CSF protein levels in the vicinity of tumor cells are superior to a uniform level, as would be expected with systemic protein. The in vivo vaccination model was designed to result in consistent gene expression and consequently focus on the effect of tumor cell expression versus nontumor cell GM-CSF expression. To confirm the in vitro results, we performed antitumor experiments in vivo using murine B16 and P815 tumor models with two GM-CSF gene therapy approaches. Previous results demonstrated that the immunization of mice ex vivo with irradiated interferon-gtransfected B16-F10.9 melanoma cells induced more potent antitumor effects than did immunization with interferon-g protein-treated tumor cells.41 We focused on replicating similar cytokine concentrations in the tumor microenvironment rather than on in vitro prestimulation and chose cutaneous transfection rather than protein injection as a method to prolong gene expression. In agreement with our in vitro results, our in vivo data demonstrated that in both B16 and P815 tumor models, mGM-CSF-transfected tumor cells had superior antitumor efficacy compared with cutaneous transfections with the mGM-CSF gene at the vaccination site. Mice vaccinated with mGM-CSF gene-transfected vaccine showed 62.5% protection against subsequent challenge with B16 tumors, whereas mice treated with nontransfected cells with cutaneous transfections of mGM-CSF showed no protection. These results on the antitumor efficacy of mGM-CSF-transfected tumor cells are consistent with our previous findings.19 We have demonstrated that inflammatory polymorphonuclear and mononuclear leukocytes are present at the site of the mGM-CSF-transfected tumor cells but not at the control DNA-transfected tumor site,19 which suggests that the transgenic mGM-CSF is functioning locally and that inflammatory cells are involved in the immunostimulatory process. The cutaneous treatment of the skin over nontransfected tumors resulted in similar GM-CSF expression in the tumor microenvironment; however, no protection from subsequent tumor challenge was noted in two different tumor systems. The in vivo vaccination model suggests that latency of expression is not a factor in the stimulation of this immune response, but that expression by the tumor cell rather than the skin is necessary. SHI, WEBER, GAN, ET AL: GM-CSF SECRETED BY cDNA TUMOR CELLS Both in vitro and in vivo results in the present study suggest that the GM-CSF produced by transfected tumor cells has a more potent immunostimulatory effect on immune cells than does the GM-CSF protein produced by other sources. Although the mechanism of this enhancement is not clear, this effect may be due either to the increased latency of hGM-CSF expression or to the transgenic mechanism of expression. However, CD81 or CD41 cells, dendritic cells, and B7-2 costimulatory molecules have been reported to be involved in the antitumor effect in GM-CSF-producing tumor cells following subcutaneous vaccination.42– 43 Whether these mechanisms are also involved in the present study remains to be investigated. In summary, the hGM-CSF cDNA delivered by particle-mediated gene transfer resulted in biologically active protein expression in cultured human tumor cell lines (MDA-MB-231 and M21). Tumor cells transfected with the cytokine gene were more effective immune stimulants than the exogenous rGM-CSF in the microenvironment of the tumor as measured by a mixed lymphocyte-tumor culture followed by cytotoxicity assay. The superior antitumor effect of tumor-secreted GMCSF was also demonstrated in vivo in B16 and P815 mouse models. Particle-mediated transfection of tumor cells is an effective technique of DNA delivery to tumor cells and may be a clinically attractive gene transfer method for human cancer gene therapy. ACKNOWLEDGMENTS We thank PowderJect Vaccines, Inc. (Madison, Wis) for use of the Accell gene gun in the study. This study was supported by a grant from University of Wisconsin Surgical Associates. REFERENCES 1. Siegel JP, Pari RK. Interleukin-2 toxicity. J Clin Oncol. 1991;9:694 –704. 2. Hank JA, Albertini M, Wesly OH, et al. Clinical and immunological effects of treatment with murine anti-CD3 monoclonal antibody along with interleukin 2 in patients with cancer. Clin Cancer Res. 1995;1:481– 491. 3. Barbuto JA, Hersh ME. Role of cytokines in cancer therapy. In: Aggarwal BB, Puri RK, eds. Human Cytokines: Their Role in Disease and Therapy. Cambridge, Mass: Blackwell Science; 1995:503–524. 4. Colombo MP, Modesti A, Parmiani G, Forni G. Local cytokine availability elicits tumor rejection and systemic immunity through granulocyte-T-lymphocyte cross-talk. Cancer Res. 1992;52:4853– 4857. 5. Golumbek PT, Azhari R, Jaffee EM, et al. Controlled release, biodegradable cytokine depots: a new approach in cancer vaccine design. Cancer Res. 1993;53:5841–5844. 6. Whiteside TL, Letessier E, Hirabayashi H, et al. Evidence for local and systemic activation of immune cells by peritumoral injections of interleukin 2 in patients with advanced squamous cell carcinoma of the head and neck. Cancer Res. 1993;53:5654 –5662. Cancer Gene Therapy, Vol 6, No 1, 1999 SHI, WEBER, GAN, ET AL: GM-CSF SECRETED BY cDNA TUMOR CELLS 7. Hank JA, Albertini MR, Schiller J, Sondel PM. Activation of multiple effector mechanisms to enhance tumor immunotherapy. J Immunol. 1993;14:329 –335. 8. Boulikas T. Cancer gene therapy and immunotherapy. Int J Oncol. 1996;9:941–954. 9. Sun WH, Burkholder JK, Sun J, et al. In vivo cytokine gene transfer by gene gun reduces tumor growth in mice. Proc Natl Acad Sci USA. 1995;92:2889 –2893. 10. Bubenik J, Voitenok NN, Kieler J, et al. Local administration of cells containing an inserted IL-2 gene and producing IL-2 inhibits growth of human tumors in Nu/Nu mice. Immunol Lett. 1988;19:279 –282. 11. Cavallo F, Dipierro F, Giovarelli M, et al. Protective and curative potential of vaccination with interleukin-2 genetransfected cells from a spontaneous mouse mammary adenocarcinoma. Cancer Res. 1993;53:5067–5070. 12. Schmidt W, Schweighoffer T, Herbst E, et al. Cancer vaccines: the interleukin-2 dosage effect. Proc Natl Acad Sci USA. 1995;92:4711– 4714. 13. Tepper RI, Pattengale PK, Leder P. Murine interleukin-4 displays potent anti-tumor activity in vivo. Cell. 1989;57: 503–512. 14. Porgador A, Tzehoval E, Katz A, et al. Interleukin 6 gene transfection into Lewis lung carcinoma tumor cells suppresses the malignant phenotype and confers immunotherapeutic competence against parental metastatic cells. Cancer Res. 1992;52:3679 –3686. 15. McBride WH, Thacker JD, Comora S, et al. Genetic modification of a murine fibrosarcoma to produce interleukin 7 stimulates host cell infiltration and tumor immunity. Cancer Res. 1992;52:3931–3937. 16. Tannenbaum CS, Wicker N, Armstrong D, et al. Cytokine and chemokine expression in tumors of mice receiving systemic therapy with IL-12. J Immunol. 1996;156:693– 699. 17. Han SK, Brody SL, Crystal RG. Suppression of in vivo tumorigenicity of human lung cancer cells by retrovirusmediated transfer of human tumor necrosis factor-a cDNA. Am J Respir Cell Mol Biol. 1994;11:270 –278. 18. Gansbacher B, Bannerji R, Daniels B, Zier K, Cronin K, Gilboa E. Retroviral vector-mediated g-interferon gene transfer into tumor cells generates potent and long-lasting antitumor immunity. Cancer Res. 1990;50:7820 –7825. 19. Mahvi DM, Burkholder JK, Turner J, et al. Particlemediated gene transfer of granulocyte-macrophage colony-stimulating factor cDNA to tumor cells: implications for a clinically relevant tumor vaccine. Hum Gene Ther. 1996;7:1535–1543. 20. Dranoff G, Jaffee E, Lazenby A, et al. Vaccination with irradiated tumor cells engineered to secrete murine granulocyte-macrophage colony-stimulating factor stimulates potent, specific, and long-lasting anti-tumor immunity. Proc Natl Acad Sci USA. 1993;90:3539 –3543. 21. Lee CT, Wu S, Ciernik IF, et al. Genetic immunotherapy of established tumor with adenovirus-murine granulocytemacrophage colony-stimulating factor. Hum Gene Ther. 1997;8:187–193. 22. Yu JS, Burwick JA, Dranoff G, Breakefield XO. Gene therapy for metastatic brain tumors by vaccination with granulocyte-macrophage colony-stimulating factor-transfected tumor cells. Hum Gene Ther. 1997;8:1065–1072. 23. Rosenberg SA. The immunotherapy and gene therapy of cancer. J Clin Oncol. 1002;10:180 –199. 24. Rosenberg SA, Anderson WF, Blaese MR, et al. Initial proposal of clinical research project: immunization of Cancer Gene Therapy, Vol 6, No 1, 1999 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 87 cancer patients using autologous cancer cells modified by insertion of the gene for interleukin 2. Hum Gene Ther. 1992;3:75–90. Cassileth PA, Podack E, Sridhar K, Savaraj N, Hanlon J. Phase I study of transfected cancer cells expressing the interleukin-2 gene product in limited stage small lung cancer. Hum Gene Ther. 1995;6:369 –383. Berns AJ, Clift S, Cohen LK, et al. Phase I study of non-replicating autologous tumor cell injections using cells prepared with or without GM-CSF gene transduction in patients with metastatic renal cell carcinoma. Hum Gene Ther. 1995;6:347–368. Cortesina G, De Stefani A, Giovarelli M, et al. Treatment of recurrent squamous cell carcinoma of the head and neck with low dose of interleukin-2 injected perilymphatically. Cancer. 1988;62:2482–2485. Miller DG, Adam MA, Miller AD. Gene transfer by retrovirus vectors occurs only in cells that are actively replicating at the time of infection. Mol Cell Biol. 1990;10: 4239 – 4242. Rakhmilevich AL, Turner J, Ford MJ, et al. Gene gunmediated skin transfection with interleukin 12 gene results in regression of established primary and metastatic murine tumors. Proc Natl Acad Sci USA. 1996;93:6291– 6296. Thompson JA, Guold MN, Burkholder JK, Yang NS. Transient promoter activity in rat mammary epithelial cells evaluated using particle bombardment gene transfer. In Vitro Cell Dev Biol. 1993;29A:165–170. Pertmer TM, Eisenbraun MD, McCabe D, Prayaga SK, Fuller DH, Haynes JR. Gene gun-based nucleic acid immunization: elicitation of humoral and cytotoxic T lymphocyte response following epidermal delivery of nanogram quantities of DNA. Vaccine. 1995;13:1427–1430. Kitamura T, Tanger T, Terasawa T, et al. Establishment and characterization of a unique human cell line that proliferates dependently on GM-CSF, IL-3, or erythropoietin. J Cell Physiol. 1989;140:323–329. Otto AM. Combinations of interferon with platinum complexes: synergistic and antagonistic effects on growth inhibition of MCF-7 and MDA-MB231 breast cancer cells. J Cancer Res Clin Oncol. 1994;120:286 –292. Helfand SC, Hank JA, Gan J, Sondel PM. Lysis of human tumor cell lines by canine complement plus monoclonal antiganglioside antibodies or natural canine xenoantibodies. Cell Immunol. 1996;167:99 –107. Ohe Y, Podack ER, Olsen KJ, et al. Combination effect of vaccination with IL2 and IL4 cDNA transfected cells on the induction of a therapeutic immune response against Lewis lung carcinoma cells. Int J Cancer. 1993;53:432– 437. Borden EC, Sondel PM. Lymphokines and cytokines as cancer treatment: immunotherapy realized. Cancer. 65: 800 – 814. Rosenthal FM, Kulmburg P, Fruh R, et al. Systemic hematological effects of granulocyte colony-stimulating factor produced by irradiated gene-transfected fibroblasts. Hum Gene Ther. 1996;7:2147–2156. Yang N-S, Burkholder JK, Roberts B, Martinell B, McCabe D. In vivo and in vitro gene transfer to mammalian somatic cells by particle bombardment. Proc Natl Acad Sci USA. 1990;87:9568 –9572. Yang N-S, Sun W. Gene gun and other non-viral approaches for cancer gene therapy. Nat Med. 1995;1:481– 483. Keller ET, Burkholder JK, Shi FS, et al. In vivo particlemediated cytokine gene transfer into canine oral mucosa and epidermis. Cancer Gene Ther. 1996;3:186 –191. 88 41. Porgador A, Brenner A, Vadai E, Feldman M, Eisenbach L. Immunization by g-IFN-treated B16 –F10.9 melanoma cells protects against metastatic spread of the parental tumor. Int J Cancer. 1991;6:54 – 60. 42. Sampson JH, Archer GE, Ashley DM, et al. Subcutaneous vaccination with irradiated, cytokine-producing tumor cells stimulates CD81 cell-mediated immunity against SHI, WEBER, GAN, ET AL: GM-CSF SECRETED BY cDNA TUMOR CELLS tumors located in the “immunologically privileged” central nervous system. Proc Natl Acad Sci USA. 1996;93:10399 – 10404. 43. Armstrong CA, Botella R, Galloway TH, et al. Antitumor effects of granulocyte-macrophage colony-stimulating factor production by melanoma cells. Cancer Res. 1996;56: 2191–2198. Cancer Gene Therapy, Vol 6, No 1, 1999