Transfection of the genes for interleukin-12 into the K1735 melanoma and the EMT6 mammary sarcoma murine cell lines reveals distinct mechanisms of antitumor activity

Abstract

Interleukin 12 (IL-12) is a pleiotropic cytokine with multiple effects on the immune system. The antitumor effects of locally produced IL-12 were examined in 2 tumor model systems. IL-12 expressing EMT6 mammary sarcomas (EMT6/IL-12) grew temporarily and then regressed resulting in mice that were immune to a further challenge of EMT6 cells. Interestingly, the IL-12 expressing K1735 melanomas (K1735/IL-12) maintained a lag phase of nonmeasurable growth for several weeks, followed by tumor outgrowth that was associated with a loss of IL-12 production. Tumor-infiltrating lymphocytes (TILs) isolated from EMT6/IL-12 tumors effectively lysed EMT6 target cells, whereas K1735/IL-12 TILs lacked lytic activity. Both IL-12 expressing tumors, however, grew progressively in nude mice indicating an important role for T cells in each case. Recombinant murine interferon gamma (rmIFN-γ) inhibited the growth of EMT6 cells, but not K1735 cells in vitro, and strongly induced the expression of the antiangiogenic chemokine interferon-inducible protein 10 (IP-10) by both cell lines. Of interest, only the EMT6 cell line was able to secrete the proangiogenic molecule, vascular endothelial growth factor (VEGF), in response to low oxygen conditions. Fluorescent staining of the vascular endothelium at the tumor injection site provided images depicting early stages of angiogenesis prior to K1735/IL-12 tumor outgrowth. These results indicate that locally produced IL-12 likely mediates the rejection of EMT6 tumors through tumor cell lysis by host immune cells, whereas its antiangiogenic potential may be counterbalanced by the strong induction of VEGF by hypoxic tumor cells. In contrast, IL-12 does not induce protective immunity to K1735 tumors. However, an antiangiogenic mechanism may be responsible for controlling tumor growth. © 2003 Wiley-Liss, Inc.

Interleukin-12 (IL-12) is a potent cytokine with pleiotropic effects on the immune system. It is a heterodimeric protein, which is produced primarily by antigen presenting cells (APC), such as macrophages and dendritic cells (DC). It plays an important role in linking the innate and adaptive immune responses. Signaling through the IL-12 receptor (IL-12R) activates both T cells and natural killer (NK) cells, triggering proliferation and the production of interferon gamma (IFN-γ).1

The efficacy of IL-12 as a therapeutic cytokine has been studied in experimental murine tumor models, and it has been used in clinical trials for the treatment of certain human cancers.2, 3, 4, 5, 6 Both systemically administered and locally produced IL-12 can be very effective in controlling tumor growth in experimental systems. How it affects tumor progression and mortality varies significantly among different tumor models,7, 8, 9 but most of the antitumor effects have been shown to be IFN-γ dependent. Neutralization of IFN-γ with antibody, for example, significantly reduces or abolishes the therapeutic effects of IL-12 in many systems.10, 11 In addition to its effects on immune cells, IFN-γ directly stimulates most somatic and malignant cells that possess a functional IFN-γ receptor. The downstream effects of IFN-γ signaling vary depending on cell type, but they may include the slowing of cell proliferation, upregulation of major histocompatibility complex (MHC) class I and class II expression and other genes involved in antigen presentation and the induction of various cytokines, chemokines and adhesion molecules.12 These observations have suggested an immune mechanism for IL-12 mediated tumor control.

In addition to its significant effects on the immune system, recent studies have also demonstrated the ability of IL-12 to inhibit angiogenesis in vivo and have shown this to be an IFN-γ dependent process as well.9, 13, 14 The precise mechanism by which IFN-γ prevents new blood vessel formation remains unclear, but involves the production of soluble antiangiogenic factors such as IP-10 (CXCL10) and monokine induced by interferon gamma (MIG) (CXCL9).15, 16 It has been suggested that the balance between the local concentrations of proangiogenic and antiangiogenic factors within a tumor dictates whether angiogenesis occurs.17, 18 Since angiogenesis is absolutely required for tumors and metastases to grow beyond a few millimeters in diameter, angiogenesis inhibition is thought to be another important mechanism by which IL-12 controls tumor growth.19, 20

Together, these studies suggest that IL-12 and IFN-γ may influence tumor growth through their interactions with a wide variety of cell types. We sought to gain a better understanding of how these different activities combine to result in tumor control or eradication. In our study, we have used gene transfection to increase the local concentration of IL-12 at the tumor site in the context of 2 different murine tumor cell lines, a melanoma and a mammary sarcoma. We have examined how this alters tumor growth, the antitumor immune response and the patterns of intratumor vascularity within each model. Here we report that the locally produced IL-12 inhibits tumor growth in these 2 model systems by distinctly different mechanisms.

CTL, cytotoxic T lymphocytes; IFN-γ, interferon gamma; IL-12, interleukin-12; IP-10, interferon-inducible protein 10; MHC, major histocompatibility complex; MIG, monokine induced by interferon gamma; NK, natural killer; rmIFN-γ, recombinant murine interferon-gamma; TILs, tumor infiltrating lymphocytes; VEGF, vascular endothelial growth factor.

MATERIAL AND METHODS

Mice and reagents

BALB/cByJ (H-2^d), mice were purchased from the Jackson Laboratory (Bar Harbor, ME), BALB/c nude (athymic) mice were from Charles River Laboratories (Wilmington, MA) and C3H/HeN (H-2^k) were from the Frederick Cancer Research Facility (Frederick MD). Hybrid (C3H/HeN x BALB/cByJ) F1 mice were bred in our animal facility at the University of Rochester vivarium. G-418 sulphate was obtained from Gibco Laboratories (Grand Island, NY), Zeocin was obtained from Invitrogen (Carlsbad, CA) and Lipofectin was obtained from BRL (Gaithersburg, MD). Paramagnetic beads (Dynabeads M-450) conjugated to sheep anti-rat-IgG were obtained from Dynal Inc. (Great Neck, NY). Heparin-sulphate conjugated agarose beads were purchased from Sigma (St. Louis, MO). Recombinant murine interleukin-12 (rmIL-12) was generously provided by the Genetics Institute (Cambridge, MA) and rmIFN-γ was obtained from R&D Systems (Minneapolis, MN).

Expression vectors, cell lines and transfectants

The expression vector pWRG 3169 encodes the p35 and p40 subunit genes for murine interleukin-12 (mIL-12), each under control of a cytomegalovirus (CMV) promoter (a gift from Dr. Hua Yu).21, 22 pHβ-Apr-1-neo contains the bacterial neo gene linked to the SV40 ori plus early promoter.23 EMT6.8 is a clone of EMT6, adapted to tissue culture from a BALB/c (H-2^d) mammary sarcoma, supplied by Dr. Sara Rockwell.24 K1735.1 is a clone of K1735, adapted to tissue culture from a C3H (H-2^k) ultraviolet radiation induced melanoma, supplied by Dr. Margaret L. Kripke.25 EMT6.8 and K1735.1 clones are referred to as EMT6 and K1735 in all experiments, respectively. Both cell lines were cotransfected with pWRG 3169 and pHβ-Apr-1-neo at a 3:1 ratio using lipofectin (BRL, Gaithersburg, MD) according to the manufacturer's instructions and selection was initiated with 400 μg/ml of G418. Clones were established by limiting dilution and were screened for IL-12 production using matched ELISA reagents for mouse IL-12 (p70) (BD Biosciences Pharmingen, San Diego, CA). YAC-1 is an H-2^a lymphoma derived from an A/Sn mouse.26 RENCA is an H-2^d renal cell carcinoma from a BALB/c mouse.27

In vitro cell growth curves

Tumor cells (1 × 10⁴) were plated in 60 mm tissue culture plates in 4 ml of media in the absence or presence of rmIFN-γ at a final concentration of 20 ng/ml. Cells were harvested at daily time points and trypan blue (Sigma) excluding cells were counted on a hemocytometer.

In vivo tumor growth and in vivo NK cell depletion

An amount of 1 × 10⁶ parental or IL-12 transfected tumor cells in a volume of 100 μl were injected intramuscularly into mice, and mean thigh diameters were determined as previously described.28 For some experiments, mice that rejected their EMT6/IL-12 tumors were rechallenged with 10⁵ EMT6 parental tumor cells in the opposite leg. Mice that were injected with K1735/IL-12 tumor cells were challenged with 10⁶ parental K1735.1 cells in the opposite leg. These secondary challenges were performed on either day 27 or day 36 after the initial injection of K1735/IL-12 cells, both occurring prior to visible outgrowth of K1735/IL-12 tumors.

Depletion of NK cells in nude mice was performed as previously described.29 Briefly, nude mice were injected intraperitoneally with either PBS or 0.5 mg of anti-IL-2 receptor beta chain (TM-β1) (provided by Dr. Richard Bankert, Roswell Park Cancer Institute, Buffalo, NY) in a volume of 0.5 ml.30 The following day mice were injected with tumor cells as above and tumor sizes were monitored. A lower dose of EMT6/IL-12 cells (2 × 10⁵) was used in this experiment to allow for slower tumor growth and a better assessment of the effects of NK cell depletion. Depletion of NK cells in the mice was assessed by staining spleen cells with a PE-conjugated anti-mouse Pan NK cell antibody (BD Biosciences Pharmingen) and analyzing them by flow cytometry (FACSCalibur; Becton Dickinson, San Jose, CA).

In vitro killing assay with tumor infiltrating lymphocytes

Tumor infiltrating lymphocytes (TILs) were purified from dissected tumors using paramagnetic beads conjugated to anti-Thy-1 antibody (T24/40.7) and were used as effector cells in a standard chromium release assay as previously described.31, 32 Target cells included parental EMT6 cells, parental K1735 cells that had been induced to express increased levels of class I MHC through incubation with rmIFN-γ at 20 ng/ml for 72 hr and the Yac-1 line, which is sensitive to NK cell mediated lysis.

Detection of IP-10 and VEGF in cell culture supernatants

Tumor cells were adapted for growth in low serum to aid in the isolation of secreted proteins. IP-10 and MIG expression was evaluated in cultured cells after incubation in the presence or absence of rmIFN-γ at a final concentration of 20 ng/ml for 24 hr. For the induction of VEGF, tumor cells were made hypoxic by plating in 60 mm glass dishes and by transferring the plates to sealed air-tight chambers connected to a manifold where air could be evacuated from each chamber and be replaced with nitrogen gas mixed with 5% CO₂. Precision gas exchanges were performed to bring the final oxygen concentration in the chambers to 1% as previously described.33 Control plates were left under normal O₂ conditions. Cell-free supernatants were collected after a 24 hr incubation period.

Heparin-binding proteins were purified from supernatants using heparin-conjugated agarose beads (Sigma). Bound proteins were eluted from the beads by boiling in Lamelli's sample buffer under reducing conditions.34 Proteins were separated on 14% SDS-PAGE gels and transferred to nitrocellulose membranes. Blots were probed with biotinylated polyclonal antibodies to murine IP-10, murine MIG or murine VEGF (R&D Systems) and binding was detected with HRP-conjugated streptavidin (Southern Biotechnology Associates, Birmingham, AL) and enhanced chemiluminesence (Amersham Pharmica Biotech, Piscataway, NJ).

Whole mount histology

Vessels within tumors were visualized using a whole mount technique recently developed in our laboratory.35 To account for intratumor heterogeneity, we examined 2–5 small (2–3 mm in diameter) pieces from different regions of each tumor. Pieces were stained in a volume of 0.2 ml with PE-conjugated antibodies against CD31 (PECAM-1) diluted to a final concentration of 4.5 μg/ml in staining buffer (PBS containing 1% bovine serum albumin and 0.1% sodium azide). Antibodies were incubated with the tumor pieces for 1 hr under moderate agitation on a rocker at 4°C. The tumor pieces were then washed in staining buffer for 1 hr on a rotator. Stained pieces were placed on microscope slides and carefully pressed between the slide and a coverslip. Slides were viewed by fluorescence microscopy and images were captured using a SONY digital camera (SONY Corporation of America, New York, NY) and Image Pro software (Version 3.0; Media Cybernetics, Silver Spring, MD). In total we have examined 4 K1735.1 tumors from days 7–14, 12 K1735/IL-12 tumors from days 7–17, 3 EMT6 tumors from days 6–7, and 5 EMT6/IL-12 tumors from days 6–14.

RESULTS

Locally produced IL-12 inhibits the growth of EMT6 and K1735 tumors

The EMT6 mammary sarcoma (H-2^d) and the K1735 melanoma (H-2^k) cell lines were transfected with a construct containing the p35 and p40 subunit genes of murine IL-12, each under a CMV promoter. We established cell lines that secreted 1–5 ng/ml of IL-12 (assessed in supernatant from 2 × 10⁵ cells grown for 48 hr in 2 ml of medium) as measured by ELISA (data not shown). To examine the ability of locally produced IL-12 to alter tumor growth in these distinct tumor models, we injected syngeneic BALB/c mice with 1 × 10⁶ IL-12-secreting EMT6 cells (EMT6/IL-12) and syngeneic C3H mice with 1 × 10⁶ IL-12-secreting K1735 cells (K1735/IL-12). Tumor growth was measured over time and compared to growth of tumors in mice injected with equal numbers of parental EMT6 or K1735 cells. Palpable tumors were detected by day 5 in mice inoculated with either parental EMT6 or K1735 tumor cell lines and the tumors grew progressively until the mice needed to be killed because of large tumor burdens (Fig. 1a and b). In contrast, mice that were injected with EMT6/IL-12 cells developed tumors that grew progressively until days 7–15 at which time all tumors began to regress (Fig. 1c). All mice given 1 × 10⁶ EMT6/IL-12 cells fully rejected their tumors and remained tumor free for several months. We rechallenged 9 mice with a lethal dose of 1 × 10⁵ parental EMT6 cells in the opposite leg 3 weeks after the initial injection of EMT6/IL-12 cells in order to test whether these mice had developed immunity to EMT6 cells. None of these mice grew tumors in either leg.

Figure 1. Locally produced IL-12 inhibits the growth of EMT6 and K1735 tumors. BALB/c mice (a,c) were injected intramuscularly with either EMT6 (a) or EMT6/IL-12 cells (c). C3H mice (b,d) were similarly injected with either K1735 (b) or K1735/IL-12 (d) cells. All mice received 1 × 10⁶ total cells in a volume of 0.1 ml. Mean thigh diameters were calculated as the geometric mean of 2 perpendicular tumor measurements at each time point. Each line represents tumor measurements for an individual mouse. Graphs are representative of numerous growth curves of mice injected with similar cell numbers.

Mice injected with K1735/IL-12 cells remained apparently tumor free for 3–4 weeks. Surprisingly, each of these mice eventually began to grow tumors with kinetics similar to parental K1735 tumors (Fig 1d). All of the mice, although apparently tumor free at day 30 (Fig. 1d), developed tumors at later time points. In a separate experiment, we observed a lag of 92 days in one mouse before the tumor rapidly grew out (data not shown). Twenty-one mice that were injected with K1735/IL-12 tumor cells were challenged with an equal number of parental K1735.1 cells in the opposite leg. These secondary challenges were performed on either day 27 or day 36 after the initial injection of K1735/IL-12 cells, both occurring prior to visible outgrowth of K1735/IL-12 tumors. Parental tumors grew in 17 of 21 mice with kinetics similar to growth in naive animals (data not shown).

We hypothesized that, during the time period of IL-12-induced growth suppression, conditions in vivo may have selected for a negative variant that no longer secreted IL-12. To determine whether the cells comprising these tumors still expressed IL-12, we removed the tumors shortly after their outgrowth was detected and placed them in culture for 3 days. Levels of IL-12 in these supernatants barely exceeded our ELISA detection limit of 500 pg/ml suggesting that only a small fraction of the cells were still secreting the cytokine. We cloned the cells from the tumors by limiting dilution in order to further confirm these results. Only 15 (12%) of the resulting 128 clones generated from 4 separate tumors still secreted detectable amounts of IL-12 as measured by ELISA.

T cells are required for the observed growth inhibitory effects of IL-12

Since IL-12 is well known as a potent stimulator of T cells and NK cells, we wanted to determine the possible roles of these cell types in each of our tumor models. We injected IL-12 secreting tumors into athymic nude mice in order to assess the role of T cells in mediating the observed growth inhibitory effects of IL-12. Nude mice that were injected with 2 × 10⁵ EMT6/IL-12 cells (Fig. 2a) or 1 × 10⁶ K1735/IL-12 cells (Fig. 2b) rapidly developed tumors that grew progressively until the mice were killed. Therefore, nude mice lacking T cells exhibited none of the IL-12 mediated growth inhibition observed in immunocompetent mice (compare Fig. 1c and d with 2a and b).

Figure 2. T cells are required for the observed growth inhibitory effects of IL-12. Nude mice were injected i.p either with PBS (a,b,e) or with anti-IL-2Rβ (c,d,f) to deplete NK cells 24 hr prior to tumor challenge. Mice were then inoculated either with 2 × 10⁵ EMT6/IL-12 cells (a,c) or 1 × 10⁶ K1735/IL-12 cells (b,d) and mean thigh diameters were calculated as in Figure 1. Spleens were removed from nude mice and NK cells were stained with an anti-mouse Pan-NK cell antibody (e,f).

We also addressed the role of NK cells in these mice by injecting an anti-IL-2Rβ antibody intraperitoneally 1 day prior to tumor cell injection. This has been shown to effectively deplete NK cells in nude mice.29 Mice that received depleting antibody prior to injection with EMT6/IL-12 tumor cells grew tumors with kinetics similar to PBS controls (compare Fig. 2a and c). The dose of EMT6/IL-12 cells (2 × 10⁵) used in this experiment allowed for slower tumor growth kinetics and a better assessment of the effects of NK cell depletion. Two of 6 mice that were treated similarly and injected with K1735/IL-12 cells grew with slightly faster kinetics than PBS controls indicating a possible minor role for NK cells in this model (compare Fig. 2b and d). We confirmed the depletion of NK cells in these mice by staining spleen cells from PBS and antibody injected mice with an anti-mouse Pan NK cell antibody. The percentage of cells staining positive with this antibody decreased from 2.7% in PBS injected mice to 0.45% in depleted mice (Fig. 2e and f).

Since the absence of T cells in nude mice abolished the growth inhibitory effects of IL-12 in both tumor models, we assessed the ability of purified TILs to recognize and kill chromium labeled tumor cell targets. Tumors were removed from BALB/c mice bearing EMT6/IL-12 tumors at a point just after the beginning of their regression phase when their mean thigh diameter measurements fell to 10–12 mm. TILs were isolated from dissociated tumor cell suspensions using paramagnetic beads conjugated to an anti-Thy-1 monoclonal antibody, which binds both T cells and NK cells. TILs from EMT6/IL-12 tumors efficiently lysed chromium labeled parental EMT6 target cells, but showed minimal activity against the NK-sensitive target Yac-1 (Fig. 3a). In contrast, TILs isolated from 10–12 mm K1735/IL-12 tumors neither recognized parental K1735 tumor cells that had been stimulated to express high levels of class I MHC, nor the Yac-1 target cells (Fig. 3b). It is important to note, however, that at the point when TILs were removed from these tumors, we were not able to detect any IL-12 secretion by the tumor cells as stated above. The cell populations present within the tumors after they started to grow out may therefore have been different from those present during the lag phase. Nonetheless, T cells were required for inhibiting tumor growth in response to IL-12 in both models (Fig. 2), but only T cells isolated from EMT6/IL-12 tumors were capable of directly killing tumor cells in vitro.

Figure 3. Recognition of parental tumor and NK-sensitive targets by tumor infiltrating lymphocytes. TILs were isolated from tumor cell suspensions using paramagnetic beads conjugated to anti-Thy-1 and cultured overnight in the presence of IL-2. These cells were used as effectors in a standard chromium release assay against parental EMT6 (▪), class I MHC induced parental K1735 (▴) or NK-sensitive Yac-1 cells (○). Data are representative of 2 experiments with TILs pooled from 2 mice per group.

IFN-γ inhibits the proliferation of EMT6 cells but not K1735 cells in vitro

IFN-γ present at high concentrations within the tumor microenvironment can have important effects on tumor cells themselves. These include direct toxicity to tumor cells and the induction of interferon-responsive genes.9, 12 In order to assess the growth response of our tumor cells to IFN-γ, we cultured each cell line in vitro in the absence or presence of exogenous recombinant (r) IFN-γ and counted the number of viable cells present on a daily basis. Both EMT6 cells and K1735 cells had similar doubling times of about 24 hr in the absence of rIFN-γ. However, when exogenous rIFN-γ was added, EMT6 cells were greatly inhibited in their ability to divide, whereas expansion of K1735 cells was relatively unaffected. Expression of IL-12 by either cell line had no effect on the in vitro sensitivity of those cells to rIFN-γ as expected (Fig. 4).

Figure 4. In vitro sensitivity of tumor cells to IFN-γ. 1 × 10⁴ tumor cells were plated in 60 mm tissue culture plates in 4 ml of media in the absence (solid bars) or presence (open bars) of rIFN-γ at a final concentration of 20 ng/ml. Cells were harvested at each time point and trypan blue excluding cells were counted on a hemocytometer. Data are plotted as averages ± 1 standard deviation of 2 experiments.

In vitro stimulation of tumor cells promotes secretion of proteins that control tumor angiogenesis

IL-12 induced IFN-γ has been shown to inhibit angiogenesis through the induction of the CXC chemokines IP-10 (CGR-2, CXCL10) and MIG (CXCL9) in certain tumor models.13, 15, 16 We examined the EMT6 and K1735 cell lines for the ability to secrete IP-10 and MIG in response to rIFN-γ using Western blots of SDS-PAGE separated cell free supernatants. RENCA tumor cells were used in this experiment as a positive control for IP-10 and MIG production in response to rIFN-γ.16 Both K1735 and EMT6 cells strongly up-regulated IP-10 secretion in response to rIFN-γ, whereas only K1735 cells expressed detectable levels of constitutive IP-10. Neither cell line secreted MIG in response to rIFN-γ (Fig. 5a).

Figure 5. In vitro stimulation of tumor cells promotes secretion of proteins that control tumor angiogenesis. (a) 1 × 10⁶ tumor cells were plated in 100 mm tissue culture plates in the presence of media only or media containing rmIFNγ at 20 ng/ml for 24 hr. (b) 5 × 10⁵ tumor cells were plated in 60 mm tissue culture plates overnight. Plates with fresh media were then incubated under normal oxygen conditions or within sealed chambers at 1% oxygen for 24 hr. (a,b) Pooled supernatant from similar plates was centrifuged to remove debris and incubated with heparin-conjugated agarose beads overnight at 4°C. Heparin-binding proteins were removed by boiling in sample buffer under reducing conditions and separated on 14% SDS-PAGE gels. Blots were probed with polyclonal antibodies to murine IP-10 and MIG (a) or murine VEGF (b). This experiment was repeated with similar results.

Ultimate control of angiogenesis, however, has been ascribed to a balance between pro- and antiangiogenic factors within the tumor microenvironment.18 Hypoxia induced VEGF has been shown to be the primary factor that stimulates angiogenesis in many tumor models.36, 37 We, therefore, assessed the ability of these cell lines to secrete the proangiogenic factor VEGF, either constitutively or in response to hypoxia. We have previously shown that EMT6 cells make small amounts of VEGF under normal oxygen conditions, but strongly up-regulate its expression under low oxygen conditions, similar to those found within hypoxic regions of EMT6 multicellular spheroids.38 These results were confirmed in our Western blots for VEGF secreted protein in EMT6 cell supernatants. Interestingly, K1735 cells failed to secrete increased levels of VEGF in response to incubation under hypoxic conditions (Fig. 5b).

Blood vessel morphology within parental and IL-12 transfected tumors

To closely examine the vascular architecture within each of our different tumors, we made use of a novel staining technique recently developed in our laboratory. Small pieces of each tumor were removed and incubated with PE-conjugated monoclonal antibodies to the endothelial cell marker CD31 (PECAM-1). After extensive washing, the stained tumor pieces were mounted on microscope slides and carefully pressed between the slide and a coverslip. This resulted in a sample that was thin enough to allow light to pass through, yet thick enough to allow for a 3D visualization of stained vessel structures using fluorescence microscopy.

EMT6 and K1735 cells responded differently towards in vitro culture conditions developed to mimic the tumor microenvironment. This suggested that the vascular architecture present within the tumors in vivo might be different between EMT6 and K1735 tumors. Vessels within EMT6 tumors 6 days after tumor cell injection appeared heterogeneous in size and irregular in shape with numerous webbed junctions characteristic of poorly formed or immature blood vessels that have not undergone remodeling (Fig 6a).39, 40 In addition, there were areas of both high and low vascular densities, confirming results previously obtained in our laboratory for EMT6 tumors using conventional histology.41 K1735 tumors, in contrast, were quite well vascularized throughout and contained long, narrow vessels with more distinct branch points than parental EMT6 tumors (Fig 6b). However, the irregular vascular structures present within both tumors were in vivid contrast to the well-organized branching patterns found in vessels within normal tissues such as muscle (Fig. 6d left side). K1735 tumors were analyzed in this experiment at 14 days after tumor cell injection in order to allow for a direct comparison with K1735/IL-12 tumors that were established but still nonmeasurable at such time. Vascular staining within these tumors did not appear significantly different at earlier time points (data not shown).

Figure 6. Vascular morphology of parental and IL-12 transfected tumors. Pieces of tumor were stained with PE-conjugated antibody against CD31 (PECAM-1) and whole mounts were viewed using fluorescence microscopy. Images are representative of analysis performed on 2–5 small (2–3 mm in diameter) pieces from different regions of each tumor. (a) EMT6 day 6 with a mean thigh diameter of 8.5 mm, (b) K1735 day 14 with a mean thigh diameter of 9.3 mm, (c) EMT6/IL-12 day 7 with a mean thigh diameter of 9.1 mm, (d) K1735/IL-12 day 15 with a mean thigh diameter of 5.7 mm and (e) higher magnification of boxed area in (d). We observed continuous sprouting vessel morphology from muscle into tumor, as seen in (d) and (e), in roughly one-third of the K1735/IL-12 tumors analyzed at early time points (days 7–15).

Vessels visualized within EMT6/IL-12 that were removed at day 7, just prior to the onset of tumor regression, were generally less webbed than those of parental EMT6 tumors that exhibited smaller vessels sprouting from larger ones. These vessels tended to be grouped together in tight bundles, and there were areas within the tumor that were nearly devoid of blood vessels directly adjacent to areas of seemingly robust angiogenesis (Fig. 6c).

Visualization of the angiogenic switch in K1735 tumors

Our in vitro data suggested that inhibition of angiogenesis might be important in preventing the early growth of K1735/IL-12 tumors, and that a switch to uninhibited angiogenesis allowed these dormant tumors to grow out when IL-12 expression was eventually lost. To visualize this angiogenic switch, we closely examined the area surrounding the injection sites of mice inoculated with K1735/IL-12 cells microscopically. This was done 15 days post injection and prior to any detectable tumor growth. We found a tiny tumor mass in a few mice that was growing lengthwise along the muscle fibers. When we analyzed the vascular staining in this area, we found contact points where the blood vessels of the surrounding muscle were sprouting branches that led into the growing tumor. These sprouting points consisted of a relatively large network of capillary-like vessels that converged into a few prominent vessels and fed into the tumor. These then began to further differentiate into the spindly shaped vessels characteristic of parental K1735 tumors (Fig. 6d and e). Therefore, the presence of IL-12 within these tumors may block this vital angiogenic transition phase resulting in tumor dormancy until the outgrowth of a nonexpressing variant in vivo.

DISCUSSION

In our study we have shown that locally produced IL-12 controls the growth of solid tumors through at least 3 distinct mechanisms. These include the stimulation of T cell mediated killing of tumor cells, suppression of tumor cell proliferation by IL-12 induced IFN-γ and the inhibition of tumor angiogenesis. Whereas previous studies have clearly identified these mechanisms in different tumor models, few have assessed the combined potential contributions of each mechanism within an individual tumor model. Our data suggests, furthermore, that the relative contributions of each mechanism in maintaining tumor control or achieving tumor eradication may be dependent on intrinsic characteristics of the particular tumors being studied.

EMT6/IL-12 tumors grew progressively for 10 days and then rapidly began to regress such that all mice fully rejected their tumors. These mice remained tumor free and were protected against a lethal challenge of parental EMT6 cells. Specific antitumor immunity mediated by cytotoxic T lymphocytes (CTL) is likely to be the primary mechanism by which these mice rejected their tumors in vivo. Our data suggests that IFN-γ produced by T cells in response to IL-12 likely stimulates EMT6 tumor cells to secrete the CXC chemokine IP-10 in vivo. However, it remains unclear whether the differences in tumor vascularity between parental EMT6 and EMT6/IL-12 tumors are due more to the direct angiostatic activity of IP-10, or to the robust immune response directed against EMT6/IL-12 tumors. IP-10 has also been shown to be a chemoattractant for activated T cells and is important in both initiating and directing T cell mediated immune responses.42, 43 The ability of the CTL to overcome the growing EMT6/IL-12 tumor was likely facilitated by enhanced recruitment of immune cells to the tumor site in response to IP-10, as well as the sensitivity of EMT6/IL-12 cells to locally produced IFN-γ, which based on the results of our in vitro studies, likely suppressed tumor cell division.

The situation with K1735 tumors was very different. K1735/IL-12 tumors displayed no outward signs of growth and persisted in a state of tumor dormancy for a variable time period ranging from 3 weeks to 92 days. The kinetics of their progression, however, matched those of recently injected parental K1735 tumors once tumor outgrowth began. Since mice injected with K1735/IL-12 tumor cells exhibited no apparent signs of antitumor immunity and IFN-γ had little effect on tumor cell division, we ascribe the growth arresting effects of IL-12 to angiogenesis inhibition in these tumors. This conclusion is further supported by the inability of these cells to make VEGF under hypoxic conditions. A recent report suggested that IL-12 induced IFN-γ can suppress the expression and activity of VEGF as well as other proangiogenic factors including matrix metalloproteinases within the tumor microenvironment.44 This may represent an additional mechanism by which IL-12 and IFN-γ maintained tumor dormancy in this model.

When we dissected the area surrounding the injection site in mice prior to tumor outgrowth, we occasionally found evidence of viable tumor growing along the muscle fibers. These tumor cells were likely feeding off of abundant muscle capillaries, which allowed their persistence during the period of angiogenesis inhibition. Recent studies suggest that during early stages of tumor development the tumor cells co-opt existing host blood vessels prior to stimulating angiogenesis.17, 45 The variable length of this dormant phase was likely a consequence of the loss of IL-12 expression being a random genetic event. Therefore, when a nonexpressing variant arose and began to proliferate, the local IL-12 and IFN-γ concentrations probably began to drop. Once the IFN-γ levels dropped below a threshold level, its inhibitory activity was likely overcome and the tumors started to grow in a fashion similar to newly injected parental K1735 tumors.

Recent studies have examined the therapeutic effects of systemic administration of rmIL-12 in mice injected with K1735 cells.9, 46, 47 For instance, Coughlin et al.46 engineered K1735 cells to express a dominant negative of the IFN-γ receptor making the cells unresponsive to exogenous IFN-γ. This study suggested that the tumor cells needed to be responsive to IFN-γ in order to mediate the growth inhibitory effects of systemic rmIL-12. Nearly all of the tumors, however, grew out once the rIL-12 therapy had ended, indicating that the mice had not developed protective antitumor immunity. In a subsequent study, Gee et al.9 continued rmIL-12 therapy of K1735 tumors for up to 3 weeks. While the growth rates of these tumors were much slower than those in untreated mice, their estimated tumor volumes still increased several fold over the course of the treatment period and showed no signs of regression. Our studies with IL-12 transfected K1735 tumor cells complement and extend these results by demonstrating long-term tumor dormancy due to IL-12 induced angiogenesis inhibition. This was likely due to the ability of the transfected tumor cells to generate a higher local concentration of IL-12, and therefore IFN-γ, than could be achieved by systemic cytokine therapy.

An important difference between our study and that of Gee et al.9 is the relative insensitivity of the K1735.1 clone to the antiproliferative and apoptosis-inducing effects of IFN-γ compared to their in vitro observations with K1735 cells (Fig. 4b). This experimental contradiction may reflect subtle phenotypic differences between the 2 cell lines. The fact that the transfectants lose IL-12 expression over time in vivo may point to an inherent genetic instability in these cells that may have produced some altered characteristics in the K1735.1 clone. Differential sensitivity to IFN-γ of variants cloned from an individual tumor cell line has been previously reported.22

The inability of IL-12 to stimulate a protective immune response to K1735 tumors probably reflects the poor relative immunogenicity of these cells. Coughlin et al.49 showed that this lack of antitumor immunity is not due to the inability of K1735 cells to be susceptible to immune mediated destruction. In this study, mice could be partially protected from tumor challenge if they were primed with K1735 cells engineered to express the T cell costimulatory molecule B7-1 (as had previously been shown48) and completely protected if IL-12 was combined with the altered cells. These results suggest that the lack of a protective response may be due to an inability to prime naive CD8+ T cells to K1735 cell tumor antigens in vivo even when IL-12 is present. We hypothesize that this may be partially due to the unavailability of tumor antigens in sufficient quantities to stimulate naive T cells while tumor growth is being repressed by angiogenesis inhibition.

Studies showing inhibition of immunity after systemic treatment with IL-12 have also been reported. Several reports, for example, have demonstrated that nitric oxide induced by IL-12 results in down-regulation of immunity.50, 51, 52 Further, a recent article has shown that administration of IL-12 to HLA.A2.1 transgenic mice specifically abrogated the immune response to HLA-A2.1 restricted epitopes in the spleen and caused a marked, but transient depletion of B cells, T cells, macrophages and DC in this organ.53 The kinetics of the transient cell depletion in their study coincides with the temporary suppression of immunity seen in IL-12 treated mice in an earlier study.8 Thus, this suppression, though temporary, can have an adverse effect on generation of immunity, and may indicate that local rather than systemic treatment may be required for the most efficacious use of IL-12. In our studies where IL-12 is secreted directly within the tumor microenvironment, we have not observed any suppression or cell depletion, suggesting local expression may be beneficial. Gene therapy offers one way to deliver cytokine locally. Recently, the feasibility of local administration, using IL-12 encapsulated biodegradable microspheres, has been demonstrated at least for tumor model systems.54

The enhanced growth kinetics of both EMT6/IL-12 and K1735/IL-12 tumors in nude mice demonstrated an essential role for T cells in inhibiting their growth. However, the roles of T cells in each case may have been different. TILs generated in response to EMT6/IL-12 tumors specifically recognize and kill EMT6 tumor cells and eventually provide immunologic memory. In contrast, TILs from growing K1735/IL-12 tumors showed no lytic activity towards K1735 tumor cells in vitro. Other studies in our laboratory have shown that both Thy-1+ and CD4+ TIL from K1735/IL-12 tumors secrete IFN-γ in vitro when incubated overnight with APC. The frequency of these cells, however, was unchanged in the presence or absence of K1735 tumor cell lysate (our unpublished observations). Therefore, the effector cells within these tumors appear to be acting in an antigen nonspecific manner. Additional studies are necessary to determine the nature of these cells. One intriguing possibility is that they may be actively recruited to the tumor by IP-10 and are further activated to produce additional IFN-γ in response to IL-12. In this way, angiogenesis inhibition may be maintained as long as IL-12 is available.

Other studies have demonstrated a prominent role for NK cells, natural killer T (NKT) cells and macrophages in response to IL-12 tumor therapy.55, 56, 57, 58 Depletion of NK cells with antibody yielded no additional increase in tumor growth kinetics in nude mice injected with EMT6/IL-12 tumors and had only marginal effects on the growth of K1735/IL-12 tumors. In addition, we were unable to isolate effector TILs from either IL-12 expressing tumor grown in normal mice that had activity against the NK-sensitive target cell line Yac-1. Whereas these results do not eliminate a potential supporting role for NK cells within immunocompetent animals, they suggest that they are not sufficient to mediate the growth inhibition mediated by IL-12 in either model. A recent study by Kito et al.58 demonstrated that macrophages stimulated by IL-12 and IL-18 had potent cytotoxic activity against murine glioma cell lines that could be blocked with inhibitors of nitric oxide. Therefore, these cells may also play an important role in the antitumor responses elicited by IL-12 within our tumor models.

A novel whole mount histologic technique developed in our laboratory allowed us to closely examine the phenotypes of blood vessels within each of our tumor models. The morphology of the vessels that infiltrate parental EMT6 and K1735 tumors appeared quite different. These results are consistent with the observation that angiogenesis proceeds through distinct mechanisms in different tissues since these tumors consist of cells of different tissue origins.39 Our results suggest that the ability of IL-12 to inhibit angiogenesis in a given tumor may depend on the phenotype of the tumor as well as the nature of the surrounding tissue. While IL-12 appeared to cause subtle differences in the formation of vessels in EMT6 tumors, it had a substantial impact on the ability of vessels to form in K1735 tumors. The loss of IL-12 expression by the K1735/IL-12 cells in vivo proved fortuitous in that it allowed us to capture images of an apparent switch from angiogenesis inhibition to the early stages of tumor vessel angiogenesis.

Our ability to detect continuous sprouting vessel morphology from muscle into tumor (Fig. 6d and e) was partially dependent on the stage of tumor growth. The images we have obtained are consistent with this vascular phenotype being a “fingerprint” of recent or ongoing angiogenesis and a marker of the angiogenic switch. Once the tumor grows beyond the immediate confines of the muscle tissue and a capsule forms, such morphology is less evident. This is likely due to continued growth and remodeling of these small branching vessels into more mature tumor vessels.39 Our results suggest that parental K1735.1 tumors rapidly progress through this stage. K1735/IL-12 tumors also initiate angiogenic sprouting from muscle tissue, but are then delayed in their progression and vascular remodeling as the local IL-12 concentration increases due to the increased numbers of IL-12 producing tumor cells. IL-12 induced angiogenesis inhibition, therefore, appears to preserve the “fingerprint” of initial muscle to tumor vascular sprouting and prevents further vascularization during the extended period of tumor dormancy.

Lee et al.47 recently demonstrated that rIL-12 could inhibit angiogenesis and growth of transplanted K1735 tumors. It had little or no effect, however, upon autochthonous virus induced mouse mammary tumors. They concluded that this difference correlated with the extent of pericyte coverage of the vessels within each tumor, where the nonpericyte covered vessels of transplanted tumors were most sensitive to the antiangiogenic activity of IL-12.47 In this respect, transplanted tumors are more similar to tumor metastases where fully transformed cells find a niche within normal tissue and rapidly induce angiogenesis. Our results suggest that rIL-12 therapy may be effective at limiting the outgrowth of tumor metastases as long as the surrounding tissue and tumor microenvironments are capable of producing the downstream effector molecules necessary for angiogenesis inhibition.

Newer studies are closely examining the complex cross talk that occurs between endothelial cells and the cells of the immune system.59 Clearly, the tumor cells themselves are heavily involved in this communication process. Experimental and clinical strategies to boost the immune response to tumors or to prevent tumor growth by inhibiting angiogenesis have met with varying degrees of success.4, 60, 61, 62 Interestingly in many cases, tumors have been shown to develop drug resistance to different angiogenesis inhibitors.63 Further studies examining the molecular phenotypes of different tumors will allow better predictions of the mechanism(s) by which various biologic agents, such as IL-12, will act in vivo. Our results demonstrate significant differences between the responses of a breast sarcoma and a melanoma to IL-12. It will be interesting to determine whether similar responses are shared among certain tumor types and whether these responses vary when tumor cells are grown in different anatomical locations. Combination therapies can then be better designed to combat each unique tumor on several fronts by influencing both the immune response and the tumor microenvironment.

Acknowledgements

Our study was funded by an NIH grant CA28332 to E.M.L. J.PM. and S.A.G. were supported by NIH training grant AI007285. C.A.M. was partially supported by the Ronald E. McNair Program.