HOXB7: a key factor for tumor-associated angiogenic switch.

We had demonstrated previously a functional bridge between altered homebox (HOX) gene expression and tumor progression through HOXB7 transactivation of basic fibroblast growth factor. Here, we have studied whether HOXB7, in addition to basic fibroblast growth factor, may induce other genes directly or indirectly related to neoangiogenesis and tumor invasion. Parental, beta-galactosidase-transduced, and HOXB7-transduced SkBr3 cell lines were examined for the expression of several growth factors and growth factor receptors involved in the proliferative and angiogenic processes. Vascular endothelial growth factor, melanoma growth-stimulatory activity/growth-related oncogenene alpha, interleukin-8, and angiopoietin-2 were up-regulated by HOXB7 transduction. The exception was angiopoietin-1 expression that was abrogated. Additional analyses included the expression levels of enzymes such as matrix metalloprotease (MMP)-2 and MMP-9 and heparanase, capable of proteolytic degradation of extracellular matrix and basement membranes. Results showed an induction of only MMP-9. The functional implication of such a finding was tested using an in vitro coculture assay in a three-dimensional matrix. A delay of differentiation with persistent nests of proliferating cells was found in endothelial cells cocultured with HOXB7-transduced SkBr3 cells. Tumorigenicity of these cells has been evaluated in vivo. Xenograft into athymic nude mice showed that SkBr3/HOXB7 cells developed tumors in mice, either irradiated or not, whereas parental SkBr3 cells did not show any tumor take unless mice were sublethally irradiated. Comparison of tumor nodules for vascularization by CD-31 and CD-34 immunostaining revealed an increased number of blood vessels in tumors expressing HOXB7. Together, the results indicate HOXB7 as a key factor up-regulating a variety of proangiogenic stimuli. Thus, HOXB7 gene or protein is a target to aim at to inhibit tumor-associated neoangiogenesis, considering the number and the redundancy of proangiogenic molecules that should be targeted one by one to theoretically achieve the same effect.

Dendritic cells infiltrating tumors cotransduced with granulocyte/macrophage colony-stimulating factor (GM-CSF) and CD40 ligand genes take up and present endogenous tumor-associated antigens, and prime naive mice for a cytotoxic T lymphocyte response.

We transduced BALB/c-derived C-26 colon carcinoma cells with granulocyte/macrophage colony-stimulating factor (GM-CSF) and CD40 ligand (CD40L) genes to favor interaction of these cells with host dendritic cells (DCs) and, therefore, cross-priming. Cotransduced cells showed reduced tumorigenicity, and tumor take was followed by regression in some mice. In vivo tumors were heavily infiltrated with DCs that were isolated, phenotyped, and tested in vitro for stimulation of tumor-specific cytotoxic T lymphocytes (CTLs). BALB/c C-26 carcinoma cells express the endogenous murine leukemia virus (MuLV) env gene as a tumor-associated antigen. This antigen is shared among solid tumors of BALB/c and C57BL/6 mice and contains two epitopes, AH-1 and KSP, recognized in the context of major histocompatibility complex class I molecules H-2Ld and H-2K(b), respectively. DCs isolated from C-26/GM/CD40L tumors grown in (BALB/c x C57BL/6)F1 mice (H-2d x b) stimulated interferon gamma production by both anti-AH-1 and KSP CTLs, whereas tumor-infiltrating DCs (TIDCs) of BALB/c mice stimulated only anti-AH-1 CTLs. Furthermore, TIDCs primed naive mice for CTL activity as early as 2 d after injection into the footpad, whereas double-transduced tumor cells required at least 5 d for priming; this difference may reflect direct DC priming versus indirect tumor cell priming. Immunohistochemical staining indicated colocalization of DCs and apoptotic bodies in the tumors. These data indicate that DCs infiltrating tumors that produce GM-CSF and CD40L can capture cellular antigens, likely through uptake of apoptotic bodies, and mature in situ to a stage suitable for antigen presentation. Thus, tumor cell-based vaccines engineered to favor the interaction with host DCs can be considered.

IgG2a induced by interleukin (IL) 12-producing tumor cell vaccines but not IgG1 induced by IL-4 vaccine is associated with the eradication of experimental metastases.

We evaluated whether antibody response correlates with tumor therapy by cytokine gene-modified tumor cell vaccines. To characterize the antibody (Ab) response against a known antigen, colon carcinoma C26 cells and C26 variants engineered to produce interleukin (IL) 12 or IL-4 were further transduced to express the human tumor-associated antigen gp38 folate receptor (FR) alpha. Irradiated IL-12- and IL-4-producing C26/FR alpha cell vaccines cured 50 and 30% of mice bearing C26/FR alpha lung micrometastases. Treatment induced a rapid, CD4-dependent Ab production dominated by IgG2a and IgG1 in response to the IL-12 or IL-4 vaccine, respectively. In contrast, untreated tumor-bearing mice showed a late serological response dominated by IgM. Anti-FR alpha IgG1 and IgG2a were able to suppress tumor metastases upon passive transfer in vivo. Sera from mice cured by the IL-12 vaccine displayed a higher binding activity, a higher anti-FR alpha IgG2a content, and a higher complement-mediated tumor cell lysis in vitro compared to the sera from nonresponder mice. Such a correlation was not found in the sera of mice treated with the IL-4 vaccine. These data indicate that cytokine-producing tumor cell vaccines strongly influence antibody response, and that in the case of the IL-12-based vaccine, the Ab titer correlates with the therapeutic response, thus suggesting its use for monitoring the outcome of vaccination in cancer patients.

Interferon gamma-independent rejection of interleukin 12-transduced carcinoma cells requires CD4+ T cells and Granulocyte/Macrophage colony-stimulating factor.

We analyzed the ability of interferon (IFN)-gamma knockout mice (GKO) to reject a colon carcinoma transduced with interleukin (IL)-12 genes (C26/IL-12). Although the absence of IFN-gamma impaired the early response and reduced the time to tumor onset in GKO mice, the overall tumor take rate was similar to that of BALB/c mice. In GKO mice, C26/IL-12 tumors had a reduced number of infiltrating leukocytes, especially CD8 and natural killer cells. Analysis of the tumor site, draining nodes, and spleens of GKO mice revealed reduced expression of IFN- inducible protein 10 and monokine induced by gamma-IFN. Despite these defects, GKO mice that rejected C26/IL-12 tumor, and mice that were primed in vivo with irradiated C26/IL-12 cells, showed the same cytotoxic T lymphocyte activity but higher production of granulocyte/macrophage colony-stimulating factor (GM-CSF) as compared with control BALB/c mice. Treatment with monoclonal antibodies against GM-CSF abrogated tumor regression in GKO but not in BALB/c mice. CD4 T lymphocytes, which proved unnecessary or suppressive during rejection of C26/IL-12 cells in BALB/c mice, were required for tumor rejection in GKO mice. CD4 T cell depletion was coupled with a decline in GM-CSF expression by lymphocytes infiltrating the tumors or in the draining nodes, and with the reduction and disappearance of granulocytes and CD8 T cells, respectively, in tumor nodules. These results suggest that GM-CSF can substitute for IFN-gamma in maintaining the CD8-polymorphonuclear leukocyte cross-talk that is a hallmark of tumor rejection.

Interleukin 12 gene therapy of MHC-negative murine melanoma metastases.

Immunological gene therapy of cancer relies heavily on the activation of T cells, but tumors with defects in MHC gene expression are not recognized by MHC-restricted T cells. To investigate the potential of cytokine genes for the therapy of MHC-negative tumors, we transduced B78H1, a class I-negative murine melanoma clone, with a polycistronic vector carrying murine interleukin (IL)-12 genes. The clones studied produced 400-25,000 pg/ml IL-12; their in vitro growth properties were similar to those of parental cells. A complete inhibition of growth was observed in vivo both after s.c. and i.v. administration of all IL-12 clones. IL-12-transduced cells were also used as a therapeutic vaccine in mice bearing micrometastases by nontransduced parental cells. A significant (80-90%) reduction in the number of lung nodules was obtained. Immunohistochemical analysis and studies in immunocompromised hosts showed that T cells and natural killer cells had a significant role in the elimination of IL-12-releasing cells. In situ hybridization with cytokine probes detected a strong increase in the proportion of leukocytes positive for IFN-gamma, tumor necrosis factor alpha, IL-1beta, and IFN-inducible protein 10 at the site of rejection of IL-12-engineered tumor cells. However, it was clear that the loss of in vivo growth was also due to T-cell- and natural killer cell-independent factors, possibly related to the antiangiogenic properties of IL-12. In conclusion, tumor therapy based on IL-12 gene transduction was effective on a MHC-negative metastatic tumor, suggesting a possible application to MHC-defective human neoplasms.

Amount of interleukin 12 available at the tumor site is critical for tumor regression.

The C26 colon carcinoma is resistant to systemic recombinant interleukin 12 (rIL-12) therapy. Transduction of C26 with genes encoding the two subunits of murine IL-12 to release 30-80 pg/ml resulted in delayed tumor onset after injection of 5 x 10(4) cells into syngeneic BALB/c mice and in 40% tumor regression after injection into CD4-depleted mice. Here, we analyzed the activity of rIL-12 (1 microgram/day) against C26 grown into CD4-depleted mice. Like in mice given injections of interleukin 12 (IL-12) gene-transduced C26 cells, depletion of CD4+ cells led to tumor regression in 6 of 14 mice, and immumocytochemical characterization of tumor-infiltrating leukocytes showed abundant infiltration by CD8+ T cells and asialoGM1+ natural killer cells, which were scanty in tumors from nondepleted mice. On the basis of the percentage of tumor regression and leukocyte infiltration we can conclude that, in the C26 system, systemic rIL-12 (1 pmicrogramg/day) produces the same results as 30-80 pg/ml IL-12 released at the tumor site. A new polycistronic retroviral vector was then used to increase the amount of IL-12 produced by C26-transduced cells. C26 cells releasing 5 ng/ml IL-12, nearly 100 times more than the above-mentioned transduced cells, were tumorigenic in less than 50% of the mice given injections of 5 x 10(4) cells. In mice given injections of 5 x 10(5) cells, an initial tumor take of 100% followed by a complete tumor regression. Tumor regression was associated with infiltration of CD8+ and asialoGM1+ cells, and mice that remained tumor free were immune to a rechallenge of nontransduced C26 cells. The results indicate that the amount of IL-12 made available at the tumor site may determine both the type and number of infiltrating leukocytes and the events leading to tumor regression as well as it may overcame host immunosuppression.

Interleukin 12 potentiates the curative effect of a vaccine based on interleukin 2-transduced tumor cells.

The purpose of these studies was to determine whether systemic administration of recombinant interleukin 12 (rIL-12) is able to potentiate an initial, but insufficient T-cell antitumor response. Mice challenged with carcinoma cells engineered to release interleukin 2 (IL-2) and displaying such a response received single or multiple i.p. injections of rIL-12. This combination of systemic rIL-12 and local IL-2 increased the percentage of mice that rejected two different IL-2 gene-transduced tumors. In another set of experiments more closely resembling a clinical situation, IL-2 gene-transduced tumors were used as vaccines in an attempt to cure mice bearing wild-type parental tumors. The combination of these vaccines with systemic rIL-12 cured mice more effectively than rIL-12 and IL-2 gene-transduced tumor vaccines alone.

Hypoxic tumor cell death and modulation of endothelial adhesion molecules in the regression of granulocyte colony-stimulating factor-transduced tumors.

C-26 colon adenocarcinoma cells transduced with the granulocyte colony-stimulating factor (G-CSF) gene form large tumors when injected into sublethally irradiated mice. These tumors regress when leukocyte function is reconstituted. Electron microscopy and immunocytochemical analysis of regressing C-26/G-CSF nodules indicates that tumor destruction is due mainly to hypoxia resulting from the functional loss of tumor vasculature and is only marginally due to direct cytolysis. Desegregation of basal lamina, cell swelling, and loss of junctions characterized the vessels within regressing tumors. Tumor cells were necrotic or filled with lipid vacuoles regardless of the distance from nearby vessels. Damage of tumor vasculature was dependent on the infiltrating leukocytes and the cytotoxic cytokines they produced. Locally produced interleukin-1 and tumor necrosis factor-alpha (TNF-alpha) induced vascular cellular adhesion molecule-1 (VCAM-1) and E-selectin on tumor vessels. Treatment with monoclonal antibodies to interferon-gamma (IFN-gamma) or TNF-alpha blocked tumor regression by inhibiting VCAM-1 and E-selectin expression on tumor-associated endothelial cells resulting in a reduced number of infiltrating leukocytes. Thus, C-26/G-CSF tumor regression presents features typical of hemorrhagic necrosis that occurs through the cytokines produced by infiltrating leukocytes in response to G-CSF.

Regression of an established tumor genetically modified to release granulocyte colony-stimulating factor requires granulocyte-T cell cooperation and T cell-produced interferon gamma.

Using the murine colon adenocarcinoma C-26 cell line, engineered to release granulocyte colony-stimulating factor (G-CSF) (C-26/G-CSF), were studied the mechanisms responsible for inhibition of tumor take in syngeneic animals and of regression of an established tumor in sublethally irradiated mice injected with these cells. Immunocytochemistry and in situ hybridization, performed to characterize tumor-infiltrating leukocytes and their cytokine expression, respectively, indicated that polymorphonuclear leukocytes (PMN) were the major cells responsible for inhibition of tumor take and that they expressed mRNA for interleukin 1 alpha (IL-1 alpha), IL-1 beta, and tumor necrosis factor alpha (TNF-alpha). Expression of interferon gamma (IFN-gamma) and of IL-4 was undetectable, consistent with the absence of T lymphocytes at the site of tumor injection. In mice injected with C-26/G-CSF cells after 600-rad irradiation, the tumors grew to approximately 1.5 cm in 30 d, regressing completely thereafter in 70-80% of mice. During the growing phase, tumors were infiltrated first by PMN (between days 15 and 20), then by macrophages, and last by T lymphocytes. Both CD4+ and CD8+ T cells were present but only CD8 depletion significantly abrogated tumor regression. Depletion of PMN by the RB6-8C5 antigranulocytes monoclonal antibody reduced the number of T cells infiltrating the tumor and prevented tumor regression. In situ hybridization performed at the beginning of tumor regression revealed the presence of mRNA for IL-1 alpha, IL-1 beta, and TNF-alpha, but also the presence of cells, with lymphoid morphology, expressing IFN-gamma. Tumors from mice treated with recombinant IFN-gamma (between days 20 and 35) were rejected faster, whereas mice treated with antibodies to IFN-gamma (from day 20) died of progressive tumor. Cyclosporin A treatment (started at day 20) also abrogated tumor regression. These results indicate that inhibition of tumor take and regression in this model occurs through different mechanisms that involve PMN and PMN-T cell interactions, respectively, as well as a combination of cytokines that, for tumor regression, require IFN-gamma. Thus, gene transfer of a single cytokine gene such as G-CSF into tumor cells appears to be sufficient to trigger the cascade of cell interactions and cytokine production necessary to destroy a cancer nodule.

Granulocyte colony-stimulating factor (G-CSF) gene transduction in murine adenocarcinoma drives neutrophil-mediated tumor inhibition in vivo. Neutrophils discriminate between G-CSF-producing and G-CSF-nonproducing tumor cells.

We have previously demonstrated that the murine colon adenocarcinoma C-26 cell line transduced with the human gene for the granulocyte CSF (G-CSF) loses tumorigenic activity through a mechanism that involved massive targeting of neutrophils at the site of tumor injection. The suppression of tumorigenicity by G-CSF was limited to the G-CSF-producing cells and was not transferred to nonproducing C-26 cells in a mixed tumor transplantation assay. We present direct evidence that neutrophils are involved in this phenomenon. We firstly examined, by electron microscopy (EM), the morphology of tumor infiltrates obtained 2, 5, and 10 days after s.c. injection of a mixture of G-CSF-producing and -nonproducing C-26 cells into syngeneic BALB/c mice. The EM analysis showed at 5, but not at 2 or 10 days, the presence of neutrophils in intimate contact with tumor cells. We then investigated whether neutrophils discriminate between G-CSF-producing and -nonproducing C-26 cells. To this aim, C-26 cells were transduced, via retroviral vector, with the Escherichia coli LacZ gene and mixed tumor transplantation assays were performed by injecting a mixture of G-CSF-producing beta-gal- and G-CSF-nonproducing beta-gal+ C-26 cells at different ratios. Histologic and EM analysis of the tumors growing at the site of injection were carried out. Five days after injection, treatment with x-gal revealed, at the histochemical level, the presence of neutrophils around G-CSF producing beta-gal- cells; cell-cell contacts and fusion of cell membranes were detected by EM only between neutrophils and G-CSF-producing cells. In vitro experiments, performed in Boyden chambers, confirmed that the G-CSF produced by C-26 cells was a chemoattractant for neutrophils. In addition, a colorimetric, cytostatic assay revealed that neutrophils were able to inhibit the growth of G-CSF-producing but not of G-CSF-nonproducing C-26 cells. Thus the tumor take after injection of G-CSF-producing C-26 cells seems to be controlled in situ through two major mechanisms namely neutrophil chemotaxis and neutrophil-mediated tumor inhibition. The results indicate that neutrophils can discriminate between G-CSF-producing and -nonproducing tumor cells and that neutrophils infiltrate the tumor mixture as long as G-CSF-producing cells are present.

Granulocyte colony-stimulating factor gene transfer suppresses tumorigenicity of a murine adenocarcinoma in vivo.

We have investigated the effect of granulocyte colony-stimulating factor (G-CSF) delivery at the site of tumor growth by transducing, via retroviral vector, the human (hu) G-CSF gene into the colon adenocarcinoma C-26 and assaying the ability of transduced cells to form tumors when injected into syngeneic mice. As a control, the same tumor cells were infected with retroviruses engineered to transduce an unrelated gene, the human nerve growth factor receptor, or carry the neomycin resistance gene only. Only cells transduced with the huG-CSF were unable to develop tumors, although huG-CSF was expressed and produced at low level as estimated by both RNA analysis and enzyme-linked immunosorbent assay, indicating that G-CSF can exert an antitumor effect at a physiological dose. Implication of G-CSF as mediator of tumor inhibition was proven by reversing the nontumorigenic phenotype of G-CSF-expressing cells with anti-huG-CSF monoclonal antibody injected at the tumor site. No tumors were formed by injecting C-26 infected cells into nu/nu mice, while neoplastic nodules appeared after injection into sublethally irradiated mice; such tumors, however, regressed when mice normalized their leukocyte counts after irradiation. Tumors were also formed after injection of a mixture of infected and uninfected C-26 cells, although critical delay in tumor formation occurred when infected cells were 10 times more represented in the mixture. Histological examination of tissues surrounding the site of injection showed infiltration of neutrophilic granulocytes, whose number correlated with that of G-CSF-expressing C-26 cells in the injected mixture. These results indicate that G-CSF may have a potent antitumoral activity when released, even at low doses, at the tumor site. The antitumoral effect is mediated by recruitment and targeting of neutrophilic granulocytes to G-CSF-releasing cells.

Adoptive immunotherapy of a BALB/c lymphoma by syngeneic anti-DBA/2 immune lymphoid cells: characterization of the effector population and evidence for the role of the host's non-T cells.

It has been previously shown that the BALB/c lymphoma YC8 is susceptible to lysis by syngeneic anti-DBA/2 lymphocytes and that YC8-bearing BALB/c mice can be cured by adoptive transfer of such immune effectors. In this study in vivo and in vitro functions of the curative immune lymphocytes have been evaluated together with the role of the host immune system in the mechanism of tumor eradication. It was found that the curative anti-DBA/2 lymphocytes were not directly cytotoxic to YC8 cells although they developed into YC8-lytic cells after in vitro restimulation with YC8. In vivo, the immune lymphocytes were able to mediate a tumor-specific delayed type hypersensitivity reaction against YC8 but had a low tumor-neutralizing activity in the Winn assay. Proliferation of infused BALB/c anti-DBA/2 lymphocytes was necessary for the in vivo therapeutic effect, since irradiation of effector cells or treatment of the donor immune lymphocytes with vinblastine abolished their curative capacity. Immunodepression of the T cell compartment of the prospective tumor-bearing animals by thymectomy plus irradiation or its abrogation in B mice (thymectomized, lethally irradiated, and reconstituted with fetal liver cells) did not interfere with the therapeutic effect of the transferred anti-DBA/2 lymphocytes. Blocking the macrophage functions of the host by carrageenan, however, abolished the therapeutic effect of immune lymphocytes. These data indicate that a radiation-resistant, non-T cell is involved in the tumor eradication induced by anti-DBA/2 lymphocytes. It was also shown that cured mice, tested 90 days after therapy, become resistant to 5 X 10(3) LD80 YC8 cells and that this resistance was due to the presence of memory cells derived from the transferred and not from the host lymphocyte population.

Alloreactivity and tumor antigens: generation of syngeneic antilymphoma killer lymphocytes by alloimmunization of mice with normal cells.

The possibility of obtaining a syngeneic antitumor effect by immunization with normal allogeneic cells was investigated by tests of the lytic activity of peritoneal exudate cells (PEC) of BALB/c mice immunized with lymphoid cells of either a single strain or a pool of six different allogeneic strains on the syngeneic Moloney virus-induced lymphoma YC8 target and on one of its clones designated YC8-D1. Significant cytotoxicity on both targets but not on two other BALB/c lymphomas was obtained with PEC of BALB/c mice singly immunized to the non-H-2-incompatible but H-2-compatible DBA/2 or B10.D2 lymphoid cells. The lack of lysis of YC8 cells by PEC of BALB/c mice immune to B10.A (H-2k,d) suggests that the in vitro killing was restricted by Kd-IEd region products of the major histocompatibility complex. Pool immunization was effective in generating antitumor cytotoxic lymphocytes only when DBA/2 lymphoid cells were included in the pool. The pattern of reactivity of effectors elicited in (BALB/c x DBA/2)F1 and in (BALB/c x B10.D2)F1 mice by immunization with DBA/2 and B10.D2 cells showed that at least two sets of antigens are recognized on YC8 targets, one shared by DBA/2 and B10.D2 tissues and the other expressed by DBA/2 cells only. Cold target blocking experiments indicated that the same effectors recognized non-H-2 antigens of DBA/2 and the cross-reacting YC8 determinants. The antitumor effect was mediated by T-cells, since it was abrogated by treatment of effectors with anti-Thy 1.2 serum plus complement. These data indicate that determinants defined by cytotoxic T-lymphocytes are expressed on the BALB/c lymphoma YC8 and cross-react with non-H-2 antigens of DBA/2 and B10.D2 strains.