Nitric Oxide Synthase Is Induced in Tumor Promoter-sensitive, but not Tumor Promoter-resistant, JB6 Mouse Epidermal Cells Cocultured with Interferon- γ-stimulated RAW 264.7 Cells: The Role of Tumor Necrosis Factor-α¹

Nitric oxide (NO), a gaseous free radical, is a short-lived molecule that has a multitude of functions in some biological phenomena(1). NO rapidly and nonenzymatically reacts with superoxide anion to form peroxynitrite anion, a highly toxic molecule causing a wide range of DNA and protein modifications (2). Conversely, there are a large number of both experimental and clinical reports showing expression of iNOS³in the tissues of chronic inflammatory diseases, including cancer. For instance, iNOS protein has been induced in premalignant and malignant,but not normal, clinical specimens from such organs as the stomach(3), colon (4), lungs (5),esophagus (6), prostate (7), and duodena(8).

A considerable fraction of the carcinogenic process involves chronic inflammation, particularly in the postinitiation stage. Upon infection with microorganisms, one of the rapid immune responses is leukocyte infiltration and activation, which produces reactive oxygen and nitrogen intermediates, prostaglandins, leukotrienes, and proinflammatory cytokines. Prolonged and excessive leukocyte activation without homeostatic regulation leads to oxidative damage to DNA and to repetitive cell death with compensatory cell division and mutation(9, 10). Thus, leukocyte activation plays a number of essential roles in the carcinogenic processes. On the other hand,little is known regarding the biochemical interactions between infiltrated leukocytes and neighboring epithelial cells, where dormant tumor cells originate and then acquire malignancy after repetitive cycles of cell death, regeneration, and tissue remodeling.

A mouse epidermal cell line, JB6, has been widely used to study tumor-promoting factors (11, 12, 13, 14, 15), antitumor promoters(16, 17, 18, 19), and the mechanisms of tumor-promoting actions(20, 21, 22). The experimental strength of this model lies in the establishment of several subclones that can be divided into two distinctive variants, transformation-sensitive (P+) subclones that undergo anchorage-independent transformation in response to tumor promoters and transformation-resistant (P−) variants that do not(23). Both P+ and P− subclones are recognized to exhibit preneoplastic phenotypes, but only the P+ cells transform into tumor cell (Tx) phenotype in response to tumor promoters, such as phorbol esters or TNF-α (23).

In the present study, we attempted to gain insight into the cellular interactions between cocultured activated mouse macrophage RAW 264.7 cells and either JB6 P+ or P− cells. As a result, iNOS protein,constitutively not detectable in either the P+ or P− subclones, was highly up-regulated in P+ but not P− subclones when cocultured with IFN-γ-stimulated RAW264.7 cells. Furthermore, this expression was mediated via macrophage-released TNF-α. The role of macrophages in epithelial iNOS expression is discussed.

DMEM and FBS were purchased from Life Technologies, Inc. (Grand Island,NY). IFN-γ was purchased from Genzyme (Cambridge, MA). All other chemicals were purchased from Wako Pure Chemical Industries Co., Ltd.(Osaka, Japan), unless specified otherwise. Recombinant mouse TNF-αand IL-1β were obtained from Chemicon International, Inc. (Temecula,CA). JB6 subclones were purchased from American Type Culture Collection. RAW 264.7 cells were kindly donated by Ohtsuka Pharmaceutical Co. Ltd. (Ohtsu, Japan).

RAW 264.7 cells (5 × 10⁵) and JB6(P+ or P−) cells (5 × 10⁵) were exclusively preincubated on a membrane culture insert (pore size, 0.45μm; diameter, 24 mm; Becton Dickinson Labware, Franklin Lakes, NJ)and a 6-well plate (diameter, 35 mm; Becton Dickinson Labware),respectively, in 5 ml of 10% FBS-supplied DMEM for 24 h. After washing the cells with PBS twice, IFN-γ (0 or 100 units/ml) was added to each cell line and incubated in 5 ml of DMEM without FBS or phenol red. After a 24-h incubation, the supernatant thus obtained was used for measuring NO₂⁻, TNF-α, or IL-1β. Residual cells were subjected to protein determination and Western blotting. These methods are described below. Each experiment was done in triplicate, and the data are mean ± SD.

A membrane culture insert, upon which RAW 264.7 cells (5 × 10⁵) were preincubated in 5 ml of 10%FBS-supplied DMEM for 24 h, was placed in a 6-well plate where JB6 P+ or P− cells (5 × 10⁵) were also preincubated under the same conditions. After washing each cell line with PBS twice, IFN-γ (0 or 100 units/ml) was added to the cells, followed by incubation in DMEM without FBS or phenol red. After a 24-h incubation, the supernatant thus obtained was used for measuring NO₂⁻. Residual cells were subjected to protein determination and Western blotting. These methods are described below. Each experiment was done in triplicate, and the data are shown as mean ± SD.

JB6 P+ cells (1 × 10⁵) were preincubated in 1 ml of 10% FBS-supplied DMEM for 24 h in a 24-well plate. Cells were then treated with 0 or 100 units/ml of IFN-γ for 12 h as negative and positive controls, respectively. Additionally, the medium from RAW264.7 cells (1 × 10⁵) on membrane culture inserts (pore size, 0.45μm; diameter, 9 mm), cultured for 12 h with IFN-γ (0 or 100 units/ml), anti-TNF-α (0 or 20 μg/ml), anti-IL-1β antibody (0 or 20 μg/ml; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and normal goat IgG (0 or 20 μg/ml; Genzyme), were added to JB6 P+ cells(1 × 10⁵) that had been preincubated for 24 h independently in a 24-well plate. After a 12 h-incubation with IFN-γ (0 or 100 units/ml) in DMEM, without FBS or phenol red, the concentrations of NO₂⁻ as well as the protein amounts were obtained as described below. Each experiment was done in triplicate, and the data are shown as mean ± SD.

A boiling lysis solution (1% SDS, 1 mm sodium vanadate,and 10 mm Tris buffer, pH 7.4) was added to the cells,which were then scraped off from the dish, sonicated, and boiled for 10 min. Ten-μg proteins were separated on 10% polyacrylamide gels and electrophoretically transferred onto polyvinylidene difluoride membranes (Millipore, MA). After blocking, the membranes were incubated with a primary antibody, either rabbit antimouse iNOS (1:1000 dilution;Affinity Bioreagents, Inc., Golden, CO) or rabbit polyclonal anti-β-actin antibody (1:1000 dilution; Biochemical Technologies,Stoughton, MA), and then with the respectively corresponding secondary antibodies, antirabbit IgG (1:1000 dilution; Dako, Glostrup, Denmark)or peroxidase-conjugated rabbit antigoat IgG (1:1000 dilution; Dako). The blots were developed using an enhanced chemiluminescence detection kit (Amersham Life Science, Buckinghamshire, United Kingdom). After each development, the antibodies were stripped, and the blots were successively reprobed with each primary antibody. The levels of iNOS,semiquantified using an NIH Image, were corrected using those ofβ-actin as an internal standard. Each experiment was done independently in duplicate twice, and the data are shown as mean ± SD.

JB6 P+ cells (1 × 10⁵) were preincubated in a 24-well plate in 1 ml of 10% FBS-supplied DMEM. After washing the cells with PBS twice, recombinant TNF-α (0, 1, 10,or 100 ng/ml), IL-1β (0, 1, 10, or 100 ng/ml), and IFN-γ (0 or 100 units/ml) were added to the cells and then incubated in DMEM, which was free of both phenol red and FBS. After 24 h, the NO₂⁻ concentrations as well as the protein amounts were measured as described below. Each experiment was done in triplicate, and the data are shown as mean ± SD.

The concentrations of NO₂⁻ in the medium were determined by a Griess assay, as reported previously(24), and the medium supernatants were subjected to the assay without dilution.

The supernatants from the medium at 0, 3, 6, 12, and 24 h after monoculture were subjected to measurement of TNF-α and IL-1βconcentrations using a commercial experimental kit (Endogen,Inc., Woburn, MA), according to the protocol of the manufacturer. The media supernatants were subjected to the assay without dilution.

Protein concentrations were determined using a DC Protein Assay kit(Bio-Rad laboratories, Hercules, CA), with BSA used as the standard.

Cytotoxicity was measured by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay,as reported previously (25).

The statistical significance of differences between groups in each assay was assessed by a Student’s t test (two-sided) that assumed unequal variance.

Stimulation of mouse RAW264.7 macrophages with IFN-γ (100 units/ml) for 24 h led to a dramatic increase in iNOS protein expression (59-fold) and NO₂⁻formation (6.4 ± 0.5 nmol/ml/mg protein, 13-fold; Fig. 1). In contrast, iNOS expression was not observed in either JB6 P+ or P−cells, with or without IFN-γ treatment. We then investigated the cellular interactions between stimulated RAW264.7 cells and either P+or P− cells separated by a transwell (pore size, 0.45 μm) in which the secreted factors and medium components can pass through, but the cells themselves cannot. We noted that IFN-γ treatment markedly induced iNOS protein expression (55-fold) in JB6 P+ but not P− cells cocultured with IFN-γ treated RAW 264.7 cells (Fig. 2,A). Accordingly, the NO₂⁻ concentration observed in the medium from the RAW/P+ coculture (30.8 ± 3.6μ m) was significantly higher than the sum of the concentrations (20.0 ± 2.3 μm;p < 0.01) added from each cell line monoculture (Fig. 2,B). On the other hand, the RAW/P−coculture generated no such significant synergistic effect (observed,18.5 ± 4.6 μm; calculated,17.9 ± 3.6 μm; Fig. 2,B). The levels of iNOS expression in RAW264.7 cells were constant after both monoculture and coculture, with either P+ or P−subclones (Figs. 1 and 2 A). No notable cytotoxicity, as detected by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay, was observed in the experiments.

IFN-γ-stimulated RAW264.7 cells are known to release a variety of proinflammatory cytokines. We have previously performed time course studies that measured the secretion of IL-1β and TNF-α from these cells after being treated with IFN-γ (100 units/ml). As shown in Fig. 3 A, IL-1β was slightly but significantly released after a 24-h incubation (0.3 ± 0.1 ng/ml/mg protein; P < 0.05), as compared with nonstimulated cells that released no detectable IL-1β over a 24-h period. In contrast, a dramatic TNF-α release was observed from 12–24 h(6.9–12.5 nmol/ml/mg protein; P < 0.05 versus control), whereas the nonstimulated cells also spontaneously released TNF-α (3.8–7.6 nmol/ml/mg protein from 12–24 h) time dependently.

To confirm that a soluble biochemical factor was released from activated RAW264.7 cells and passed through the transwell to induce iNOS expression in P+ cells, we added conditioned media from 12-h-stimulated P+ cells or RAW264.7 cells to the P+ subclones and then incubated them with IFN-γ for another 12 h. As shown in Fig. 4, neither conditioned medium from IFN-γ-stimulated-P+ cells nor that from nonstimulated RAW 264.7 cells showed any significant increase in NO₂⁻ concentration in the media of JB6 P+ subclones, as compared with that observed in the media of P+cells stimulated by IFN-γ for 24 h. In contrast, the 12-h-IFN-γ-stimulated RAW 264.7 cell-conditioned medium notably enhanced NO₂⁻ generation 2.2-fold (P < 0.001). Furthermore, we believe that it is extremely important to note that the specific anti-TNF-α antibody, but not the anti-IL1-β antibody or nonspecific IgG, diminished NO₂⁻ formation by 84% (P < 0.001).

Treatment of JB6 P+ cells with IFN-γ (100 units/ml) for 24 h led to NO₂⁻ accumulation(1.2 ± 0.2 nmol/ml/mg protein; Fig. 5). The recombinant TNF-α (1–100 ng/ml), together with IFN-γ,markedly enhanced NO₂⁻production 1.8–3.8-fold (P < 0.05) as compared with the positive control, whereas IL-1β (1–100 ng/ml)showed no significant synergy with IFN-γ. On the other hand, TNF-α(1–100 ng/ml), but not IL-1β, also potentiated NO₂⁻ production by 1.5–6.3-fold over the negative control (P < 0.05) in the absence of IFN-γ. The enhancing effects caused by a combination of TNF-α and IL-β, with or without IFN-γ, on NO₂⁻ formation were almost comparable with those by TNF-α alone. No notable cytotoxicity was observed in the experiments.

The present study provides experimental evidence that IFN-γ-stimulated RAW264.7 cells secrete TNF-α via a paracrine loop to induce iNOS expression in JB6 P+ but not P− cells. Although not in a statistically significant manner, the RAW 264.7 cell-conditioned medium with added PBS slightly enhanced NO₂⁻ production (Fig. 4). We assume that this was attributable to the effect of the spontaneously released TNF-α (Fig. 3,B). The TNF-α concentration after a 12-h stimulation with IFN-γ (1.4 ± 0.3 ng/ml; Fig. 3,B) was within the range of that in medium of a separate experiment that had recombinant TNF-α added (1–100 ng/ml; Fig. 5). On the other hand, the possibility of iNOS protein induction by TNF-αvia an autocrine loop from P+ cells can be ruled out, because the stimulated P+ cell-conditioned medium did not significantly enhance iNOS induction. IFN-γ synergistically enhanced the levels of iNOS induction by TNF-α. A similar synergism between IFN-γ and TNF-αhas been observed previously by Chan et al.(26) and other groups. We also investigated the role of another proinflammatory cytokine, IL-1β, with iNOS expression in P+cells, because it was demonstrated previously to induce iNOS expression in murine lung epithelial cells with a combination of both IFN-γ and TNF-α (27). However, our present experiments with antibody neutralization (Fig. 4) and recombinant IL-1β addition (Fig. 5) excluded the involvement of IL-1β, even as a synergist, from the iNOS-expressing pathways in P+ cells. iNOS protein was expressed in JB6 P+ cells when cocultured not only with IFN-γ-stimulated RAW 264.7 cells but also with those treated with lipopolysaccharide alone and a combination of IFN-γ and lipopolysaccharide.⁴

TNF-α has been reported to induce the activation of an iNOS gene transcriptional factor, NF-κB, via the function of an endosomal acidic sphingomyelinase, the product of which, ceramide, induces degradation of an NF-κB inhibitor (28). The iNOS induction by TNF-α may also be mediated via the Jun NH₂-terminal kinase and stress-activated protein kinase signal transduction pathways (29). In addition,TNF-α induces transformation of JB6 cells via the Jun NH₂-terminal kinase (30) and NF-κB pathways (21). COX-2 is another essential inducer of inflammation, and we have found recently that the expression level of COX-2 protein in JB6 P+ but not P− cells increased by 2-fold as compared with the control when cocultured with IFN-γ-stimulated RAW264.7 cells.⁴ Therefore, the dual induction of iNOS and COX-2 by TNF-α might play some important roles in the transformation pathways of JB6 cells. The observed lack of NF-κB activity in P−subclones⁵ can provide a rationale for the contrastive susceptibility between P+ and P− cells to the TNF-α-induced iNOS expression. On the other hand, it is notable that TNF-α and/or IFN-γ did not induce iNOS expression in promoter-resistant P−variants,⁴ suggesting that cytokine-induced iNOS expression is selective for the epithelial cells in which transformation-related, transcriptional factors, such as activator protein-1 or NF-κB, are active.

Thus far, noticeable differences in cellular phenotypes between the JB6 P+ and P− subclones have been reported to be involved with the activity of such transcriptional factors as activator protein-1(31, 32, 33, 34), arachidonic acid metabolism (35),oxidation sensitivity (36), expression of the tissue inhibitor of matrix metalloproteinases (37, 38, 39, 40), and phospholipid metabolism (41). The present study shows, for the first time, that there is a substantial difference between these two subclones in their susceptibility to TNF-α-induced iNOS expression.

Studies with immunohistochemical staining to this point have localized iNOS expression to epithelial cells (3, 4, 5, 6, 7, 8, 42),monocytes/macrophages (5, 43, 44, 45), joints(46), and endothelial and smooth muscle cells(47). Although certain bacteria are known to directly induce epithelial iNOS expression in vitro (48, 49), an alternative route for inducing epithelial iNOS expression through interactions between inflammatory leukocytes and epithelial cells might be plausible. The present results lead to the hypothesis that NF-κB-active epithelial cells themselves express iNOS protein using neighboring macrophage-secreted TNF-α. Although activated macrophage-induced epithelial iNOS expression remains to be demonstrated in vivo, some coculture studies have shown iNOS expression in colon epithelial cells with T lymphocytes(50), airway epithelial cells with lung mononuclear cells(51), and retinal pigment epithelial cells with activated T cells (52). These findings may support our hypothesis regarding the mechanisms of epithelial iNOS expression. It is important to stress that the above-mentioned three reports did not specifically address which cellular properties are a prerequisite for iNOS expression. As described above, whereas both JB6 P+ and P− subclones exhibit a preneoplastic phenotype, the presence or absence of NF-κB activity readily differentiates these two variants. In fact, the constitutive activation of NF-κB is one of the conspicuous cellular properties involved with malignancy and tumorigenicity (53, 54). Along a similar line, a coculture of macrophages with the L929 fibrosarcoma cell line showed synergistic NO generation(55), further suggesting that a neoplastic phenotype is required for epithelial iNOS expression.

The process of clonal expansion of initiated cells requires time,because it involves repetitive cell death, regeneration, and the remodeling of neighboring normal cells. During these repeated steps,inflammation plays a pivotal role in many organs of potential carcinogenesis, such as the stomach and colon. Upon infection with organisms like a virus or bacteria, one of the rapid immune responses is the infiltration and activation of monocytes/macrophages and neutrophils. Neutrophil-mediated inflammatory phenomena have been reported to be the hydroxylation, nitration, and chlorination of protein (56) and DNA (57) by reactive oxygen and nitrogen intermediates, which induce alterations of cellular properties including the acquisition of malignant phenotypes. Although these biochemical modifications may be relevant to the carcinogenic process, it is possible to imagine that neutrophil-derived free radicals have much shorter life spans, limiting their possibility of access to the target molecules of epithelial cells through the extracellular matrix. With this in mind, it can be supposed that iNOS expression in neoplastic cells may cause more severe oxidative injuries to themselves than by intercellular oxidative insult, i.e.,endogenously generated NO and its more reactive metabolites, such as peroxynitrite, would rapidly and effectively react with the intracellular components.

In conclusion, the present results lead us to hypothesize that NF-κB-active epithelial cells induce endogenous iNOS expression through TNF-α that is secreted via a paracrine loop from activated macrophages. Moreover, this event may lead to more efficient and substantial oxidative injuries to the cells themselves.