The role of microenvironment interleukin 1 (IL-1) on 3-methylcholanthrene (3-MCA)–induced carcinogenesis was assessed in IL-1–deficient mice, i.e., IL-1β−/−, IL-1α−/−, IL-1α/β−/− (double knockout), and mice deficient in the naturally occurring inhibitor of IL-1, the IL-1 receptor antagonist (IL-1Ra). Tumors developed in all wild-type (WT) mice, whereas in IL-1β–deficient mice, tumors developed slower and only in some of the mice. In IL-1Ra–deficient mice, tumor development was the most rapid. Tumor incidence was similar in WT and IL-1α–deficient mice. Histologic analyses revealed fibrotic structures forming a capsule surrounding droplets of the carcinogen in olive oil, resembling foreign body–like granulomas, which appeared 10 days after injection of 3-MCA and persisted until the development of local tumors. A sparse leukocyte infiltrate was found at the site of carcinogen injection in IL-1β–deficient mice, whereas in IL-1Ra–deficient mice, a dense neutrophilic infiltrate was observed. Treatment of IL-1Ra–deficient mice with recombinant IL-1Ra but not with an inhibitor of tumor necrosis factor abrogated the early leukocytic infiltrate. The late leukocyte infiltrate (day 70), which was dominated by macrophages, was also apparent in WT and IL-1α–deficient mice, but was nearly absent in IL-1β–deficient mice. Fibrosarcoma cell lines, established from 3-MCA–induced tumors from IL-1Ra–deficient mice, were more aggressive and metastatic than lines from WT mice; cell lines from IL-1–deficient mice were the least invasive. These observations show the crucial role of microenvironment-derived IL-1β, rather than IL-1α, in chemical carcinogenesis and in determining the invasive potential of malignant cells. [Cancer Res 2007;67(3):1062–71]
- chemical-induced carcinogenesis
The development of malignant changes in association with chronic inflammation has long been recognized in several clinical conditions and is supported by animal tumor models (reviewed in refs. 1– 9). Persistent expression of proinflammatory cytokines, in or near tumors, likely exerts pleiotropic effects, ranging from increasing growth and invasiveness of the malignant cells to activation of immune-mediated mechanisms, leading to the destruction of tumor cells and inhibition of tumor growth. Of special relevance to the process of inflammation and malignant transformation are interleukin 1 (IL-1) and tumor necrosis factor α (TNFα). These two proinflammatory cytokines are considered “alarm” cytokines, as they are generally not expressed in health, but are synthesized by macrophages soon after confronting the inflammatory insult. IL-1 and TNFα further activate stromal cells and infiltrating leukocytes to potentiate and sustain the local inflammatory response (reviewed in refs. 10, 11).
Within the IL-1 family, IL-1β and IL-1α are prominent agonists mediating inflammatory and immunomodulatory effects (reviewed in refs. 10– 16). The IL-1 receptor antagonist (IL-1Ra), also a member of the IL-1 family, binds to the IL-1 type I receptor and specifically prevents either IL-1β or IL-1α from triggering a signal (reviewed in refs. 10– 16). Although recombinant IL-1β and IL-1α bind to the same receptors and exert the same array of biological functions, they differ in the compartments in which they are active in vivo. In order to be active, IL-1β requires cleavage of its inactive precursor by caspase-1, followed by release of the mature molecules from macrophages and other cells. In contrast, IL-1α is active either as an intracellular molecule or as an integral membrane form and is only rarely secreted by cells, mainly macrophages, following processing of the precursor by calpain (reviewed in refs. 10, 11, 17). We have assessed the differential roles of tumor cell–derived IL-1α compared with tumor-derived IL-1β on tumor-host interactions, using fibrosarcoma cell lines overexpressing either of the active forms of the IL-1 molecules. The expression of cell-associated IL-1α by tumor cells increases their immunogenicity by activating effective antitumor responses that lead to tumor regression and the establishment of an immune memory that protects mice from the malignant WT cells ( 18– 22). In contrast, fibrosarcoma cells that actively secrete IL-1β are more aggressive than the virulent parental cells, likely due to an IL-1β–induced cascade of inflammatory mediators, resulting in increased tumor angiogenesis, enhanced invasiveness, as well as tumor-mediated immunologic suppression ( 23, 24). This was also substantiated in other experimental tumor systems ( 25, 26). In support of this concept, IL-1β−/− mice fail to develop B16 melanoma tumors, due to the absence of host-derived IL-1β required for angiogenesis and invasiveness ( 20). In several human cancers, local IL-1 expression by the malignant cells or the microenvironment has been associated with aggressive tumor growth and poor prognosis (reviewed in ref. 11). Treatment of mice with the IL-1Ra or the concomitant overexpression of IL-1Ra by the malignant cells was shown to inhibit the growth and metastasis of human cancer xenografts which express IL-1, although not affecting their in vitro growth rate ( 27, 28). This indicates that IL-1 is essential for the invasiveness of malignant cells.
Most studies on the effects of proinflammatory cytokines on malignant processes have assessed the tumorigenicity of existing malignant cells or the invasiveness of tumors. However, studies on the effect of proinflammatory cytokines in the process of carcinogenesis have only recently been initiated. For example, microenvironment-derived TNFα was studied for the development of malignancies of the skin ( 29– 32) and liver carcinogenesis ( 33) in TNFα/TNF receptor–deficient mice, and in other models of carcinogenesis promoted by chronic inflammation ( 34, 35). To our knowledge, there are no direct studies on the role of the different IL-1 molecules in the process of local carcinogenesis. In the present study, we have assessed the differential involvement of IL-1 in chemically induced carcinogenesis, injecting 3-methylcholanthrene (3-MCA) into WT or knockout mice that lack genes of the IL-1 family, i.e., IL-1α−/−, IL-1β−/−, IL-1α/β−/− (double knockout mice), and IL-1Ra−/− mice. The effect of host-derived IL-1 on the cellular characteristics of the arising malignant cells was also assessed.
Materials and Methods
Mice. Female BALB/c mice were purchased from Harlan (Jerusalem, Israel). The generation of IL-1 knockout mice, i.e., IL-1α−/−, IL-1β−/− and IL-1α/β−/− (double knockout mice), and IL-1Ra−/− mice was previously described ( 36). These strains of mice are homozygous for the relevant mutation. The IL-1/IL-1Ra−/− mice were bred and kept at the Animal Facilities of the Faculty of Health Sciences, Ben-Gurion University (Beer-Sheva, Israel), under aseptic conditions. Mice were treated according to the Animal Care NIH guidelines adapted by our Animal Committee.
Induction of 3-MCA–induced tumors. Mice were injected s.c. into the right thigh with 3-MCA (Sigma Israel, Rehovot, Israel) dissolved in olive oil (200 μg/mouse; ref. 37). In this experimental system, local fibrosarcomas developed within 3 to 5 months. Mice were inspected twice a week for tumor development. When tumors reached a diameter of 10 mm, mice were sacrificed and the tumor tissue was aseptically removed. Part of the tissue was immediately fixed in formalin for histologic analyses, and the rest of the tissue was processed for the establishment of cell lines by enzymatic digestion in trypsin (10 min at 37°C).
Invasiveness of 3-MCA–induced cell lines. For determining tumorigenicity patterns of 3-MCA–induced cell lines from IL-1/IL-1Ra−/− and WT mice, 2 × 105 fibrosarcoma cells were injected intrafootpad into BALB/c mice. Tumor development was assessed twice a week using a caliper. For experimental metastasis, 2 × 105 cells were injected i.v. into the tail vein. After 10 days, mice were sacrificed and the lungs were removed. Metastasis load was evidenced by macroscopic evaluations and by the weight of the lungs.
Immunohistochemistry. Samples from the site of 3-MCA injection were obtained on days 10 and 70, fixed in 4% paraformaldehyde, dehydrated in alcohol, cleared in xylene, and embedded in paraffin. Four-micron sections were stained with H&E using established protocols. For immunohistochemistry, tissue sections were deparaffinized in xylene and rehydrated with decreasing concentrations of alcohol. Endogenous peroxide was blocked with hydrogen peroxide and antigen retrieval was done by 0.01 mol/L of sodium citrate (pH 6.0) for 1 min in a pressure cooker. After blocking in the appropriate normal serum, tissue sections were stained with primary antibodies. Antibodies were used as follows: goat polyclonal anti-mouse vimentin (1:100; Santa Cruz Biotechnology, Santa Cruz, CA), rat anti-mouse F4/80 (1:20; Serotec, Oxford, United Kingdom), purified rat anti-mouse Gr-1 (1:20; eBioscience, San Diego, CA), rabbit polyclonal anti–Von Willebrand factor (1:200; Dako, Cytomatrin, Glostrup, Denmark), mouse monoclonal antibodies anti–proliferating cell nuclear antigen (PCNA, 1:100; Dako), goat anti-mouse IL-1β (1:20; R&D Systems, Minneapolis, MN), mouse polyclonal anti-COX-2 (1:1,000; Cayman Chemical Company, Ann Arbor, MI). Vectastain Elite ABC Peroxidase kit (Vector Laboratories, Inc., Burlingame, CA) was used for secondary antibody application and detection. Visualization was done by using AEC as a substrate (Zymed Laboratories, Inc., San Francisco, CA). A pathologist examined the slides in a blind manner.
In vivo treatment with inhibitors of inflammation. Mice were treated with inhibitors of inflammation 2 days before carcinogen injection and then everyday for 10 days, thereafter, mice were sacrificed and tissue from the site of 3-MCA injection was removed and processed for histologic analysis. The following inhibitors were used: IL-1Ra (50 mg/kg, i.p.), TNF binding protein (TNF BP), a recombinant construction of soluble extracellular domains of human TNF p55 receptor (10 mg/kg, i.p.; ref. 38; both reagents were kindly provided by Amgen, Thousand Oaks, CA) and the cyclooxygenase inhibitor Abitren (75 mg/kg, i.m.; ABIC, Netanya, Israel).
Statistical analyses. Each experiment was repeated at least three to five times with similar patterns of responses. In vivo experiments of 3-MCA–induced tumorigenicity consisted of 10 to 20 mice in each experimental group, whereas in experiments on the invasiveness of 3-MCA–induced fibrosarcoma cell lines, experimental groups consisted of 5 to 10 mice. The data shown are from pooled or single representative experiments, as indicated, and are expressed as mean values ± SE. Significant differences in results were determined using two-sided Student's t test; P < 0.05 was considered significant.
3-MCA–induced carcinogenesis is impaired in mice deficient in IL-1β, but is enhanced in mice deficient in IL-1Ra. The carcinogen 3-MCA serves as an initiator and promoter of tumorigenesis; tumors with the characteristic of fibrosarcomas develop 3 to 5 months after a single injection. 3-MCA–induced carcinogenesis was assessed in WT BALB/c, IL-1β−/−, IL-1α−/−, IL-1α/β−/− (double knockout) mice, and in mice that are deficient in the IL-1Ra. As shown in Fig. 1 , the most rapid tumor development was observed in IL-1Ra−/− mice. In contrast, in mice deficient in IL-1β, i.e., in IL-1β−/− and IL-1α/β−/− mice, impaired tumorigenicity patterns were observed, as manifested by the incidence of tumors as well as the delay in tumor development. Tumorigenicity patterns in WT BALB/c and IL-1α−/− mice were similar and showed an intermediate phenotype. For example, on day 120 after carcinogen injection, in 100% of IL-1Ra−/− mice and 50% of WT and IL-1α−/− mice tumors were apparent, whereas a much lower tumor incidence was observed in IL-1β–deficient mice (12% for IL-1β−/− and 28% for IL-1α/β−/− mice, respectively).
From these data, IL-1β clearly plays a crucial role in the development of 3-MCA–induced tumors, because in the absence of IL-1β expression, there is a markedly reduced onset and incidence of tumors. In contrast, in mice deficient in the naturally occurring IL-1Ra, enhanced tumor development was observed. The congenital absence of IL-1Ra results in a condition in which IL-1β or IL-1α can bind to their receptors without competition from their naturally occurring antagonist. The development of spontaneous rheumatoid arthritis in these IL-1Ra–deficient mice has been reported ( 39– 42). As shown in Fig. 1, deficiency in IL-1α did not impair 3-MCA–induced carcinogenesis compared with WT mice. These results reveal distinct effects of microenvironment-derived IL-1α and IL-1β in carcinogenesis.
The early leukocyte infiltrate at the site of 3-MCA injection in IL-1/IL-1Ra–deficient mice. To evaluate the nature of the local tissue response to the carcinogen, tissue sections from the site of 3-MCA injection were obtained on day 10. As the carcinogen is solubilized in olive oil, tissue sections revealed multiple lipid droplets, each encapsulated by a thin layer of connective tissue. These encapsulated droplets were observed in each of the IL-1 family deficiency groups as well as in WT mice. However, as shown in Fig. 2 , there were significant differences in the intensity of the leukocytic infiltrate surrounding the carcinogen-containing lipid droplets.
In IL-1Ra−/− mice, a dense leukocytic infiltrate surrounded the encapsulated droplets and this infiltrate was also prominent in the surrounding connective tissues and muscles ( Fig. 2E). The infiltrate consisted mainly of neutrophils as well as scattered macrophages, which had the appearance of foam-like cells. Almost no lymphocytes were observed in any of the infiltrates. Some encapsulated droplets were heavily invaded by neutrophils, displaying the morphology of abscesses, within which necrotic tissue was observed. Some infiltrated droplets in the IL-1Ra−/− mice had morphologies similar to that of lipid or foreign body granulomas.
In BALB/c mice, a connective tissue capsule consisting of fibroblasts and matrix fibers surrounded the droplets of the carcinogen; however, the leukocytic infiltrate was dramatically reduced compared with that observed in IL-1Ra−/− mice ( Fig. 2A and E). In WT mice, the leukocytic infiltrate was sparse, localized mainly around the carcinogen-containing droplets and did not invade the surrounding tissues. The infiltrate that was observed in IL-1α−/− mice was similar to that of WT mice ( Fig. 2B). In IL-1β−/− ( Fig. 2C) and IL-1α/β−/− ( Fig. 2D) mice, the leukocyte infiltrate was almost absent.
In each IL-1 family–deficient group, mice were injected with olive oil, without the carcinogen. In all cases, droplets of lipid surrounded by a thin layer of connective tissue were present but without remarkable leukocyte infiltrates, even in IL-1Ra mice (results not shown).
The cellular composition of the site of 3-MCA injection on day 10 was further characterized by immunohistochemical analyses in tissue sections from three prototypic experimental groups, i.e., WT, IL-1Ra−/−, and IL-1α/β−/− mice. As shown in Fig. 3 , the fibroblast content was clearly more pronounced in IL-1α/β−/− than in WT mice and was lowest in IL-1Ra−/− mice, as manifested histologically with anti–vimentin antibodies, a marker of fibroblasts. This is possibly due to differences in infiltrating inflammatory cells that invade the tissue. When infiltrating neutrophils were visualized by anti–Gr-1 antibodies, a markedly dense infiltrate was observed in sections from IL-1Ra−/− mice, whereas only a few neutrophils were observed in sections from WT mice, and there were almost no detectable neutrophils in sections from IL-1α/β−/− mice. Using macrophage-specific anti-F4/80 antibodies, a sparse infiltrate was observed at the site of 3-MCA injection in all experimental groups. In summary, it seems that in IL-1Ra−/− mice, a dense neutrophilic infiltrate surrounds the fibrotic, encapsulated lipid droplets containing the carcinogen, whereas in IL-1α/β−/− mice, a fibrotic response, without inflammation was prominent. In WT mice, an intermediate inflammatory response was observed.
To further characterize the tissue responses at the site of carcinogen injection, proliferating cells were identified with anti-PCNA antibodies. A pronounced proliferative response was observed in tissues of IL-1Ra−/− mice, as early as 10 days after the injection of the carcinogen. The proliferating cells were distributed around the encapsulated lipid droplets and in the leukocytic infiltrate. In tissues from WT and IL-1α/β−/− mice, there were fewer PCNA-positive cells, which were more numerous around the encapsulated lipid droplets. Patterns of staining in IL-1α−/− mice were similar to those in WT mice and stainings in IL-1β−/− mice were similar to those in IL-1α/β−/− mice (results not shown).
At the site of carcinogen injection, fibrotic and inflammatory responses were observed. The proliferating cells observed here may thus include fibroblasts or cellular components of the inflammatory responses, i.e., stromal cells or endothelial cells.
Host-derived IL-1 induces the early leukocyte infiltrate at the site of 3-MCA injection in IL-1Ra–deficient mice. In order to assess the role of microenvironment-derived IL-1 in the induction of local leukocytic infiltrate at the site of 3-MCA injection, IL-1Ra−/− mice, in which the early leukocyte infiltrate was most pronounced, were treated with different inhibitors of inflammation. Thus, after the injection of 3-MCA, IL-1Ra−/− mice were treated daily with recombinant IL-1Ra, TNF BP, or Abitren, a cyclooxygenase inhibitor. On day 10, mice were sacrificed and tissue samples from the site of 3-MCA injection were histologically analyzed for the local inflammatory response. As can be seen in Fig. 4 , treatment with IL-1Ra strongly inhibited leukocyte infiltration, whereas a more modest response was observed with Abitren. No effect was observed following treatment with TNF BP. The same patterns of changes in inflammatory responses were also observed in immunohistochemical studies assessing neutrophils, IL-1β, COX-2 ( Fig. 4), IL-1α, and TNFα (results not shown). Similar results were obtained in 3-MCA–injected WT mice, however, as the inflammatory response in WT mice was very modest on day 10, the results were not shown. These inflammatory responses were local, at the site of carcinogen injection, and no systemic manifestations were detected. These results indicate that under our experimental conditions, microenvironment-derived IL-1, rather than TNFα, induces the early inflammatory response at the site of 3-MCA injection.
Late changes in inflammatory responses at the site of 3-MCA injection. The leukocyte infiltrate at the site of 3-MCA injection was also assessed on day 70 and is depicted in Fig. 2F–J. Compared with day 10, there is a pronounced increase in the intensity of the leukocytic infiltrate in WT ( Fig. 2F) and IL-1α−/− mice on day 70 ( Fig. 2G), in which the local infiltrate was dense and had already invaded the muscle. In addition, atypical fibroblastoid-like cells were abundant. In contrast, at sites of 3-MCA injection in IL-1β−/− and IL-1α/β−/− mice, the leukocytic infiltrate was still sparse, although a little increased compared with that observed on day 10 ( Fig. 2H and I, respectively). In sections from IL-1Ra−/− mice, the leukocytic infiltrate was denser, compared with day 10, and malignant cells were present and widespread in the tissue, although overt tumors were not yet evident ( Fig. 2J). These data indicate that tumors arise from sites of the encapsulated droplets of 3-MCA, in which inflammatory responses are particularly abundant in IL-1Ra−/− mice.
We further characterized fibroblasts, neutrophils, macrophages, proliferating cells, and blood vessels at the site of 3-MCA injection on day 70. As depicted in Fig. 5 , fibroblasts stained with anti–vimentin antibodies were widely distributed in the tissue surrounding the 3-MCA–containing droplets in IL-1Ra−/− mice, but to a lesser extent in WT mice. On the other hand, in IL-1α/β−/− mice, fibroblasts were mostly located in the area of the lipid droplets rather than in the surrounding tissue. Immunohistochemical staining with anti-PCNA antibodies revealed that in IL-1Ra−/− and in WT BALB/c mice, proliferating cells were widespread in the tissue around carcinogen-containing droplets, whereas almost no proliferating cells were observed in sections from IL-1α/β−/− mice. Similar patterns were observed with anti–Gr-1 antibodies characterizing neutrophils. On day 70, the neutrophilic infiltrate in IL-1Ra−/− mice was less dense than on day 10. It should be noted that in WT mice, the neutrophilic infiltrate appeared later (day 70) than in IL-1Ra−/− mice (day 10). In IL-1α/β−/− mice, neutrophils were almost absent in sections from the site of 3-MCA injection, even on day 70. In stainings with anti-F4/80 antibodies, we observed a dense infiltrate of macrophages in sections from IL-1Ra−/− and WT mice, and to a much lesser degree in IL-1α/β−/− mice. Thus, the premalignant or malignant cells are embedded in tissue with an intense inflammatory reaction during the process of carcinogenesis. Angiogenesis was assessed with antibodies against Von Willebrand factor, a marker of endothelial cells. Pronounced angiogenesis was observed in tissues from IL-1Ra−/− mice, as manifested by the abundance of organized blood vessels and scattered endothelial cells. In tissues from WT mice, the angiogenic response was evident, although less pronounced than in IL-1Ra−/− mice, with smaller and less organized blood vessels. Almost no angiogenesis was observed on day 70 in sections from IL-1β (results not shown) or IL-1α/β–deficient mice. Thus, signs of malignancy, such as proliferating cells and blood vessel formation, were observed at the site of 3-MCA injection; they correlated with the local inflammatory response and with the timing and incidence of tumor development.
3-MCA–induced fibrosarcoma cell lines from IL-1/IL-1Ra–deficient mice exhibit differential invasiveness. Cell lines obtained from 3-MCA–induced tumors were established and their potential for in vivo invasiveness was characterized following intrafootpad injection into WT mice. These fibrosarcoma cell lines specifically lack IL-1 or IL-1Ra genes because they were generated from tumors in deficient mice. As can be seen in Fig. 6A , tumor cell lines generated from IL-1Ra−/− mice exhibited a higher degree of invasiveness as compared with malignant cells derived from WT mice or IL-1α/β−/− mice that were the least invasive. This increased degree of invasiveness was also evident at the level of lung experimental metastasis following i.v. injection of the malignant cells. Lung metastases was apparent both by gross morphology of the lungs ( Fig. 6B) and by an increase in the lung weight ( Fig. 6C), the latter due to metastatic load. Lung metastases induced by a 3-MCA–derived cell line from IL-1Ra−/− mice was increased compared with a cell line that was induced in WT mice. Cell lines from IL-1α/β−/− mice were almost devoid of metastatic potential (results not shown). This behavior of the cell lines was consistent in the vast majority of cell lines tested ( 5– 10) from each phenotype.
The present study shows that microenvironment-derived IL-1β plays a pivotal role in the process of 3-MCA–induced chemical carcinogenesis. This carcinogen acts as a complete carcinogen, mediating the functions of an initiator and promoter of tumor development. Thus, in mice devoid of IL-1β, i.e., in IL-1β−/− and IL-1α/β−/− (double knockout) mice, 3-MCA–induced tumors developed in only 40% to 60% of the treated mice, respectively. Equally important, in IL-1β–deficient mice with tumors, the malignant process developed after a long lag phase (∼100–120 days), compared with 50% tumor incidence in the same time period in WT mice. The patterns of tumor development in mice deficient in IL-1α were the same as in WT mice. In contrast, in IL-1Ra−/− mice, tumor development was the most rapid (100% of the mice developed overt tumors by day 120), compared with 60% of WT or IL-1α−/− mice, but no tumors in IL-1β−/− or IL-1α/β−/− mice. IL-1Ra reduces the biological activity of both IL-1β and IL-1α (reviewed in refs. 10– 15, 43). However, the activity of IL-1α is mainly cell-associated, particularly active as an integral, biologically active membrane cytokine ( 10, 11, 17).
In promoting chemical-induced carcinogenesis, IL-1α was less effective than IL-1β, as its activity was mainly cell-associated. For IL-1–driven inflammatory responses, the secreted form of IL-1 (mostly IL-β) is essential, as it acts on stromal and infiltrating cells, to induce the production of a cascade of inflammatory molecules/cytokines and adhesion molecules that sustain and expand the inflammatory response. The secretion of IL-1β in BALB/c mice, and particularly, in mice deficient in IL-1Ra, results in a greater number of infiltrating neutrophils to the site of the injected carcinogen. The intensity of the inflammatory response likely reflects chemokine production, which may be a result of secreted IL-1β rather than membrane IL-1α. To our knowledge, this is the first report that directly shows the role of IL-1 in chemical carcinogenesis and attributes differential roles to microenvironment-derived IL-1β and IL-1α in tumorigenesis.
The mechanism of action of IL-1β, and to a lesser extent, IL-1α, in the process of chemical-induced carcinogenesis are certainly multiple, as IL-1 is a pleiotropic cytokine, affecting each stage of the malignant process (reviewed in refs. 10, 11). In the process of carcinogenesis, IL-1 is possibly responsible for both shortening the time of tumor development and increasing the rate of tumors, by amplifying the action of the carcinogen via the intense inflammatory responses it induces. IL-1 may affect the initial stages of carcinogenesis, in synergy with the initiating activities of 3-MCA, to induce mutations in the local tissue fibroblasts, mainly by activating neutrophils, macrophages, and stromal cells to produce mutagenic oxygen and nitrogen intermediates. It was previously shown that the addition of phorbol 12-myristate 13-acetate–activated neutrophils to fibroblasts in culture resulted in their transformation, due to the activation of the oxidative respiratory burst ( 44). Microenvironment-derived IL-1 may further affect tumor promotion by activating NF-κB in the premalignant cells, thus rescuing them from apoptosis, and further enabling their proliferation, via stimulating growth factor production in stromal or infiltrating cells. Extensive proliferation may further lead to the accumulation of additional mutations, causing overt transformation of the target cells. By stimulating local inflammatory responses, IL-1 may further contribute to tumor angiogenesis, invasiveness, dissemination, and metastasis. Indeed, we have previously shown the cardinal role of host- and tumor cell–derived IL-1β, rather than IL-1α, in tumor invasiveness ( 20, 23, 24).
Our results show that host-derived IL-1, rather than TNFα, mediates the early (day 10) inflammatory response at the site of carcinogen injection, as neutrophilic infiltration was inhibited by IL-1Ra, but not by TNF BP. In addition, local TNFα levels, at the site of carcinogen injection, did not differ dramatically between control or IL-1/IL-1Ra–deficient mice (results not shown). Although IL-1 and TNFα synergize in inducing inflammatory responses, it is possible that under different experimental conditions, unique IL-1 or TNFα-dependent inflammatory responses are activated. For example, Miller et al. recently showed the critical role of IL-1 signaling in the recruitment of leukocytes to infection sites ( 45). By using a combination of TLR-2, MyD88, and IL-1RtI–deficient mice, it was shown that IL-1 signaling, rather than TLR-2 signaling, was essential for the eradication of Staphylococcus aureus by infiltrating neutrophils. However, other studies have emphasized the role of TNFα-induced inflammatory responses in experimental carcinogenesis ( 5, 34, 35, 46).
We purposefully studied the severity of the inflammatory response at the injection site of the carcinogen. A thin layer of fibrotic tissue containing fibroblasts, ECM fibers, and other stromal elements were consistently observed and encapsulated the droplets of olive oil in which the carcinogen was dissolved. These encapsulated droplets, generated by the local fibrotic response, were apparent in all groups of mice, as early as 10 days after injection of 3-MCA and they persisted in the tissue until tumors developed; in fact, tumors seemed to develop from the area of encapsulated carcinogen droplets. The differences in the tissue response, at the site of carcinogen injection, were in the local inflammatory response, with the appearance of granulomas, infiltrated by neutrophils and some macrophages, which correlated with the levels of IL-1β expression. Blankenstein's group recently described a similar mechanism of carcinogen encapsulation by foreign body–like granulomas which consist of fibroblasts that form a scar tissue around droplets of the carcinogen, thus inhibiting diffusion of the carcinogen into the tissue ( 47, 48). This possibly diminishes tissue injury, protects against DNA damage, and ultimately inhibits tumor development. Indeed, protective effects of such granulomas were experimentally shown in mice injected with low doses of 3-MCA, where low tumor incidence was observed. In IL-1Ra−/− mice, products of the marked inflammatory response may impair the structure of the granuloma, enable the diffusion of the carcinogen into the tissue and also enhance tumorigenesis by the mechanisms described above. Indeed, in tissue sections from IL-1Ra−/− mice, a heavily infiltrated reactive granulation tissue, characterized by extensive cell proliferation, possibly of fibroblasts or cellular elements of the inflammatory response, as well as neoangiogenesis, was already observed on day 10 after 3-MCA injection. It is noteworthy that fibroblasts are the target cells for transformation in this experimental system and their extensive proliferation throughout the process of carcinogenesis may contribute to transformation.
Massive macrophage infiltration was observed at the site of 3-MCA injection on day 70 in IL-1Ra−/− and WT mice and to a lesser degree in IL-1α/β−/− mice. On day 50, the macrophage infiltrate was still sparse (results not shown). Thus, at early stages of carcinogenesis, neutrophils were dominant in the leukocytic infiltrate at the site of 3-MCA injection, however, at later time intervals, macrophages were prevalent in the infiltrate. Thus, it seems that in our experimental system, neutrophils and their products are important for cell transformation, whereas macrophages appear possibly at a time when premalignant and even malignant cells already exist and they promote their proliferation and invasiveness. The protumorigenic characteristics of tumor-associated macrophages have been thoroughly described (reviewed in ref. 3).
In our studies, no significant infiltration of lymphocytes was observed in 3-MCA–induced granulomas, indicating that this process is mediated by innate immunity mechanisms. IL-1 induced by 3-MCA may serve as the stimulus for the inflammatory response observed around carcinogen droplets. As IL-1Ra−/− mice spontaneously develop rheumatoid arthritis in a process that is T cell–dependent ( 39– 42), it needs to be determined whether inflammatory responses that precede tumor development in IL-1Ra−/− mice stem from high levels of unopposed IL-1, or whether activated lymphocytes and/or their products also contribute to the initiation of the inflammatory response in a remote manner. It was recently shown that T cells ( 49), B cells, and their secreted antibodies ( 50) were essential for the initiation of the innate cell inflammation that is involved in tumorigenesis in certain experimental systems via recruitment of the innate inflammatory infiltrate.
Imprints of the “IL-1 milieu”, in which 3-MCA–induced tumors developed, were also shown by the increased invasiveness of transplantable fibrosarcoma cell lines that were established from tumors that developed in WT and especially in IL-1/IL-1Ra–deficient mice. Such tumor cell lines were subsequently injected into control syngeneic BALB/c mice in order to assess their invasiveness. Thus, 3-MCA–induced tumor cells lines from IL-1Ra−/− mice are significantly more invasive, metastatic, and induce more potent angiogenic responses (results not shown), as compared with tumor cell lines recovered from WT mice, whereas lines from IL-1α/β−/− mice were the least invasive. The increased invasive patterns of cell lines derived from IL-1Ra−/− mice may stem from inherent genetic properties of the malignant cells that were imprinted in the cells by IL-1–dependent events, during the process of carcinogenesis, i.e., potent chronic inflammatory responses. In cell lines from IL-1Ra−/− mice, compared with WT or IL-1α/β−/− mice, we could not detect differences in polyploidity of cells in karyotype analyses or in the expression of specific H-ras mutations, which are abundant in 3-MCA–induced fibrosarcomas (results not shown). Although not mutually exclusive, it is possible that potent and unopposed IL-1 activity in cell lines generated from IL-1Ra−/− mice induce IL-1–dependent local inflammatory responses, which are essential for the invasiveness of the malignant cells. These possibly involve IL-1 secretion by the malignant cells, constitutively or following stimulation by local inflammatory signals, and the subsequent activation of stromal cells and the recruitment and activation of leukocytes, to produce an IL-1–induced cytokine/proinflammatory mediator cascade, including IL-1–induced IL-1 secretion. In immunohistochemistry studies, in tumor sections, we have observed IL-1 expression both in the malignant cells and in cellular elements of the microenvironment. Also, supernatants of cell lines derived from 3-MCA–induced IL-1Ra−/− mice contained more chemokines/cytokines, and manifested a stronger angiogenic potential than supernatants of cell lines derived from WT or IL-1α/β−/− mice (results not shown). These results are in accordance with our previous observations on the role of both malignant cell- and host-derived IL-1β in promoting tumor invasiveness ( 20, 23, 24). The present findings show the correlation between the development of a heavy and persistent leukocytic infiltrate, initially consisting of neutrophils and later of macrophages, with the high incidence of tumor development in a relatively short period; this was particularly manifested in IL-1Ra−/− mice. Accordingly, in WT and IL-1α−/− mice, in which the inflammatory response develops later, tumors appear late. In mice lacking IL-1β, in which the inflammatory response at the site of carcinogen injection was minimal, with no apparent granulomas, the incidence of tumor development was lower, however, a significant percentage of the mice developed tumors. This latter observation suggests that tumor development occurs in the presence of a low but inflammatory response at the site of tumorigenesis. In contrast, when the carcinogen activates an intense chronic local inflammatory response, it enhances tumorigenesis by shortening the time of tumor development, increasing tumor incidence, and by promoting the generation of more malignant tumor cell variants.
In conclusion, microenvironment-derived IL-1α and IL-1β have different effects in the process of chemical-induced carcinogenesis as well as in determining the characteristics of malignant cells that would subsequently develop. Further understanding of the role of the IL-1 molecules in the process of chemical carcinogenesis and tumor invasiveness may lead to the development of novel chemoprevention approaches, based on the neutralization of specific IL-1 molecules.
Grant support: The Israel Cancer Association, the Israel Ministry of Health Chief Scientist's Office, and the Concern Foundation (E. Voronov); the Israel Ministry of Science in cooperation with the Deutsches Krebsforschungscentrum (Heidelberg, Germany), the United States–Israel Bi-national Foundation, the Israel Science Foundation founded by the Israel Academy of Sciences and Humanities, the Israel Ministry of Health Chief Scientist's Office, the Association for International Cancer Research, and the German-Israeli DIP collaborative program (R.N. Apte); and NIH grants AI-15614, HL-68743, and CA-04 6934 (C.A. Dinarello).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The authors thank Rosalyn M. White, Dr. Lubov Gayvoronsky, and Natalya Pasternak for their help.
Note: Y. Krelin and E. Voronov contributed equally to this work.
Prof. Ron N. Apte is an incumbent of the Irving Isaac Sklar Chair in Endocrinology and Cancer.
S. Segal, deceased February 26, 2006.
- Received August 9, 2006.
- Revision received November 13, 2006.
- Accepted November 28, 2006.
- ©2007 American Association for Cancer Research.