Induction of Iron-derived EPR Signals in Murine Cancers by Nitric Oxide EVIDENCE FOR MULTIPLE INTRACELLULAR TARGETS*

The cell-mediated immune response to syngeneic tu- mors activates the cytokine-inducible nitric oxide synthase. We observed that syngeneic murine tumors exhib- ited EPR signals related to iron-nitrosyl complex formation. Three different EPR active iron-nitrosyl species were observed, an Fe(RS)2(N0)2 signal and two dif- ferentiable heme-nitrosyl complexes. Hemoglobin as- says showed that the heme-nitrosyl signals were not derived from contaminating hemoglobin. Signal ampli-tudes were attenuated in mice treated with N"-mono- methyl-L-arginine (MU), an inhibitor of nitric oxide synthase. Tumors grown in vivo contained EPR signals while those grown in culture without continuing cytokine stimulation lost the signals after a few days. Cul- tured cells that were treated with cytokines, or that were cocultivated with cytokine-activated macrophages, regained EPR active complexes. These results show that the cell-mediated immune response to syngeneic tumors involves the induction of nitric oxide syn- thase. While nitric oxide synthesis is induced in both tumor infiltrating

** Visiting faculty member from the Department of Internal Medicine, Hematology/Oncology, of the Chonbuk National University Medical School, Chonbuk, Korea.
for their catalytic function. Further studies demonstrated that enzyme inhibition involved the [4Fe-4Sl prosthetic group rather than the apoenzyme (10). It was also observed that tumor cells cocultivated with activated macrophages released a significant fraction of their intracellular iron in parallel with development of inhibition of DNA synthesis and mitochondrial respiration (5, 8,10,11).
NO biosynthesis is inhibited by several N-substituted L-arginine derivatives, including, Nu-monomethyl-L-arginine (MLA)l (1) but not by the respective D-arginine stereoisomers, thus providing a useful experimental tool for demonstrating which effector functions of the cell mediated immune response are mediated by the L-arginine:NO pathway (1, 2). Commoner et al. (12) found that a g = 2.039 EPR signal appeared in the livers of rats which were fed various chemical carcinogens. This signal was due to formation of a paramagnetic Fe-NO-thiol complex in the tissues (13). Vanin et al. (14,15) and Butler et al. (16) structurally characterized inorganic, EPR active Fe.RS.NO complexes and found them to have a general chemical formula of Fe(RS)2(N0)2. The complexes were found to be unstable with either oxidation or polymerization leading to loss of the EPR signal. This iron-based EPR signal was recently demonstrated in activated macrophages and their target cells after the induction of the high output NO synthase (17)(18)(19). Identification of Fe(RS)2(N0)2 complex formation in tumor cells links activated macrophage-induced inhibition of iron-containing enzymes with nitric oxide biosynthesis.
Heme-nitrosyl EPR signals have been described in tumors from various sources. Brennen et al. (20) described a three-line signal found in two murine tumors, a reticulum cell sarcoma and a neuroblastoma. Subsequent papers by other groups (21) described a similar signal in several other tumors. Normal tissue was found to not contain the three-line signal, although, Maruyama et al. (22) described such a signal in normal tissues held at room temperature for 1-2 days. This three-line signal was subsequently identified as being caused by nitrosylated heme-containing proteins (23)(24)(25)(26)(27).
We describe here evidence that both the Fe(RS)2(N0)2 and the heme-NO EPR signals are products of the cell-mediated immune response to tumor cells in vivo. Furthermore, these signals can be induced in cultured tumor cells by cytokines or activated macrophages in vitro. Reagents-Murine interferon y (IFNy) and tumor necrosis factor (Y (TNF) were obtained as a generous gift from Genentech (San Francisco, CA). Interleukin-la (IL-1) was a gift from Hoffman La Roche.

MATERIALS AND METHODS
i%mor Cell Lines-A number of murine tumor cell lines were used i n these experiments, including RD-995, LR-351, B10.2, MetWA, MCA 106, S-180, and EMT-6. RD-995 and LR-351 were ultraviolet radiationinduced spindle cell tumors arising in C3WHeN mice. RD-995 is termed a "progressor" tumor because it can grow in normal mice following subcutaneous implantation of 1-5 x lo6 cells (28). In contrast, LR-351, a "regressor" tumor, can grow only in UV radiation-exposed or y-irradiated mice, and is rejected by normal animals. B10.2 is a spindle cell tumor induced in B10 mice by methylcholanthrene (gift from D. Lynch, Immunex Corp., Seattle, WA). Meth/A (gift from Lloyd Old, Memorial Sloan Kettering Cancer Center, New York) and MCA 106 (gift from Benjamin Kim, University of Utah) are methylcholanthrene-induced spindle cell tumors arising in BALBlc and C57BU6N mice, respectively (29). S-180 (American Type Culture Collection, Rockville, MD) is a spontaneously arising sarcoma originally derived from a Swiss Webster mouse which will also grow in C3HLHeN mice (30). EMT-6 cells are a tained from Dr. Robert Kallman, Stanford University (Palo Alto, CAI. spontaneous BALB/c mammary adenocarcinoma cell line, originally ob-EMT-6 cells were grown and maintained as described (31). Other tumor cells were maintained by serial subcutaneous passage in syngeneic mice or by culture in RPMI 1640 medium supplemented with 5% fetal calf serum (Hyclone Laboratories, Inc., Logan, UT), 100 unitdml glutamine (Sigma), 100 units of penicillin G (Sigma), and 50 pg/ml streptomycin (working medium). Tumor cells were released from culture flasks by rinsing monolayers with isotonic phosphate-buffered saline (PBS, pH 7.4) followed by brief treatment (5-10 min) with a trypsin-EDTA solution (0.25% trypsin from hog pancreas (490 USP unitdml, ICN Biomedicals, Cleveland, OH), 0. Nitrite Measurement-Biologically produced NO is rapidly oxidized to nitrite and nitrate in aqueous solutions (33,34). Nitrite concentration in the cell-free culture supernatants, therefore, served a s a reflection of nitric oxide production and was measured by a modification of the colorimetric Griess reaction (35). Briefly, 50-pl aliquots of the culture supernatants dispensed into 96-well microtiter plates (flat bottom) (Coming Glass Works, Coming, N Y ) were incubated with 100 pl of a 1:l mixture of 1% sulfanilamide (Sigma) in 30% acetic acid and 0.1% N-(l-naphthy1)ethylenediamine dihydrochloride in 60% acetic acid at room temperature. After 20 min, absorbance was measured a t 570 nm using a Dynatech MR 5000R plate reader (Dynatech Laboratories, Inc., Alexandria, VA). Concentrations were determined from a linear standard curve obtained from serial concentrations (6.25-100 p~) of sodium nitrite (Sigma) in working medium. Results of triplicate measurements were presented as mean 2 S.D.
Microtiter Hemoglobin Assay-The possibility that in vivo tumor samples were contaminated by erythrocyte-derived hemoglobin was evaluated by using a modification of the standard colorimetric assay for hemoglobin. In brief, tumor cell suspensions were resuspended a t lo8 celldml 1% Triton X-100. Avolume of tumor cell lysate (50 pl) and serial dilutions of human hemoglobin (20-1000 pg/ml) in 50 pl volume were added to microtiter wells along with 50 pl of Drabkin's reagent (12 m~ NaHCO,, 0.76 m~ KCN, 0.61 m~ K,Fe(CN)6). After 10 min, absorbance due to cyanomethemoglobin was measured at 570 nm (36). The sensitivity of this assay was 1 pg of hemoglobirdwell (310 m).
Macrophages-Peritoneal exudate cells were obtained from mice injected with 2 ml of sterile 3% thioglycollate (Difco Laboratories, Detroit, MI) (32). Four days after injection, Peritoneal exudate cells were harvested by peritoneal lavage with 6 ml of ice-cold PBSmouse. Peritoneal exudate cells were washed twice and then resuspended a t 2 x lo6 celldml in working medium. After a 2-h incubation in 75-cm2 tissue culture flasks (a final volume of 20 muflask) (Becton Dickinson Labware, Oxnard, CA) at 37 "C in humidified 95% air, 5% CO, atmosphere, nonadherent cells were removed by repeated washing with ice-cold PBS. Adherent cells (peritoneal exudate macrophage) were then used for further experiments.
Macrophage a n d RD-995 Cocultures-Peritonea1 exudate macrophages were cultured a t 2 x lo6 celldml in the presence or absence of recombinant mouse IFNy (25 unitdml, Amgen Biologicals, Thousand Oaks, CA) and lipopolysaccharide (100 ng/ml, from Escherichia coli serotype 0128:B12, Sigma). After an 18-h incubation a t 37 "C in a 5% C 0 2 incubator, peritoneal exudate macrophages were harvested using a cell scraper (Baxter S/PR Brand diSPoR cell scraper, Baxter Healthcare Corporation, McGraw Park, IL). Following extensive washes, activated or unstimulated macrophages were added to confluent RD-995 cultures in 75-cm2 culture flasks (5 x lo6 macrophages in a final volume of 10 ml of working medium). After a 48-h culture in the presence or absence of 500 p~ MLA at 37 "C in a 5% CO, incubator, cell-free culture supernatants were assayed for nitrite. Cells were harvested using the trypsin-EDTA solution and analyzed by EPR spectroscopy. EPR Spectroscopy-After dispensing thick cell suspensions into quartz EPR tubes (Wilmad Glass Co., Buena, NJ), the tubes were subjected to a centrifugation at 500 x g for 5 min. The volume of the packed cells was controlled to maintain 140 putube. The samples were stored at -70 "C until spectra were recorded. EPR spectra were recorded on a Bruker ER-2OOD spectrometer operating at x-band with instrument settings a s follows: microwave frequency, 9.4 GHz; microwave power, 1.7 milliwatts; modulation frequency, 100 kHz; and modulation amplitude, 1.0 millitesla. During data collection, the temperature was maintained at 77 K by immersing the samples in liquid nitrogen in a quartz insert dewar. The field was calibrated periodically with a sample of reduced methyl viologen. Instrument gains are stated in the appropriate figure legends.
Peak heights for Fe(RS)2(N0)2 EPR signals were measured from the top of the g = 2.039 peak to the inflection point at g = 2.027 while heme-nitrosyl EPR signals were quantitated by measuring the difference between the base line and the peak at g = 2.078.

RESULTS
Identification of EPR Signals in lhmor Tissue-Freshly dissociated RD-995 tumor cells contained four different EPR signals (Fig. 1, spectrum A ). The dominant species in the spectrum was an axial signal with peaks atg = 2.039 and 2.012 indicative of formation of an iron-dithiol-dinitrosyl complex of the general formula, Fe(RS)2(N0)2. A second species was produced by a heme-nitrosyl moiety. This signal which is dominated by three equally spaced lines centered a t g = 2.012 with a hyperfine coupling constant of 1.7 millitesla is produced by five coordinate heme-iron with NO bound at the axial position (26,27). A third species seen in freshly dissociated RD-995 cells has principal g values at 2.078 and 1.988 and derives from a hemenitrosyl moiety in which the iron is six coordinate, that is, with the proximal histidine still bound to the iron. EPR signals in proteins containing a six coordinate heme-nitrosyl signal are easily converted in vitro to the five coordinate complex by addition of denaturing agents such as SDS or urea (23). The fourth signal seen in RD-995 cells is an isotropic signal, located a t g = 2.003 and derives mainly from semiquinone radicals produced either by normal cell metabolism or as an artifact of sample preparation (37). Comparison of EPR Signals Generated in lhmors in the Presence and Absence of MU-Since the cell-mediated immune response to tumors may activate the cytokine-inducible nitric oxide synthase, we investigated whether tumors growing in mice exhibited EPR signals related to Fe(RS)2(N0)2 complex formation. 3.2 x lo8 freshly dissociated RD-995 cells contained Fe(RS)2(N0)2 and Heme-NO EPR signals with relative peak heights of 174 and 58 units, respectively (Fig. 1, spectrum A ) . Freshly dissociated RD-995 cells from MLA-treated animals exhibited a 5.5-fold decrease in the amount of nitrite detected in the medium. Similarly, the Fe(RS)2(N0)2 and heme-nitrosyl EPR signals (Fig. 1, spectrum B ) had peak heights of 45 and 31 units, respectively, in 3.2 x 10' cells. These peak heights correspond to 3.9-and 1.9-fold decreases in the amount of Fe(RS)2(N0)2 and heme-nitrosyl complexes produced in the MLA-treated cells relative to untreated cells.

Comparison of EPR Signals from lhmors Obtained in Vivo and in V i t r e T o determine whether the apparent in vivo nitrosylation was related to host cell infiltration of tumors, EPR
signals generated from tumor cell lines cultured in vitro were compared with signals derived from the same line growing in vivo. While all of the tumors obtained in vivo contained one or more NO-derived EPR signals (Table I), those grown in vitro without cytokine stimulation contained only the semiquinone radical signal as illustrated in Fig. 1 Fig. 2, showed that nitrite production by the cells correlated well with Fe(RS)2(N0)2 EPR signal height, suggesting that the signals are related to induction of endogenous nitric oxide synthesis. Unlike macrophages, which do not respond to IL-1 (38, 39), induction of tumor cell NO synthesis was more potently induced by IL-lLFNy than by T N F d F N y . A heme-nitrosyl EPR signal also appeared in the IFNyl TNFa-treated cells. The size of the heme-nitrosyl signal did not appear to change in cells treated either with IFNyAL-1 or with all three cytokines (Fig. 21, while the size of the Fe(RS)2(N0)2 signal increased substantially, especially in cells treated with all three cytokines (Fig. 2). This suggests that NO is bound to most of the accessible heme in the cells before the formation of Fe(RS)2(N0)2 complexes begins.
Effect of Cytokine Levels on Iron-Nitrosyl Signals in Cultured EMT-6 Cells-Murine EMT-6 adenocarcinoma cells grown in culture for multiple passages over a period of several years were exposed to cytokines in concentrations that varied over several orders of magnitude. Four of six control cultures exhibited a small heme-nitrosyl EPR signal and a very small Fe(RS)2(N0)2 signal (Fig. 3, spectra A and B ) . Control cells which contained EPR signals also contained low levels of nitrite and nitrate (Fig. 4) while medium from the two cultures not exhibiting an EPR signal did not contain detectable levels of either nitrate or nitrite. The Fe(RS)2(N0)2 EPR signal tripled in amplitude in EMT-6 cells treated with low levels of cytokines (0.02 unitdml IFNy, 0.02 unitdm1 IL-1, 0.04 unitdml TNFa) while the heme-nitrosyl signal approximately doubled in size (Fig. 3, spectrum C ) . The nitrite and nitrate levels in the culture medium of these cells also increased slightly (Fig. 4). Nitrite and nitrate levels in the culture medium continued to increase as the cytokine levels were increased (Fig. 4). Fe(RS)2(N0)2 EPR signal heights also increased with increasing cytokine levels, reaching a maximum then dropping off slightly at higher cytokine levels (Fig. 4). The heme-nitrosyl signal on the other hand did not increase in size beyond that seen in samples containing the lowest levels of cytokines (Fig.  4). These results suggest that a heme compound(s) is the first target of cytokine induced NO with formation of Fe(RS)2(NO)2 complexes occurring at higher NO levels.
EMT-6 cells grown subcutaneously in BALB/c mice develop heme-nitrosyl EPR signals (data not shown). When tumors are removed from the animal and grown in culture, they continue n m o r Iron-Nitrosyl EPR Signals Induced by NO to exhibit predominantly heme-nitrosyl EPR signals unless the cells are treated with cytokines. Cytokine treatment induces a strong Fe(RS)2(N0)2 signal which nearly obscures the hemenitrosyl signal. These results suggest that the EMT-6 cells grown in vivo are exposed to relatively low levels of NO.
Hemoglobin Assays in lhmor Cultures-Several possibilities exist concerning the cellular origin of the heme-nitrosyl signal in freshly isolated tumor cells. One possible explanation was that this signal was due to nitrosylation of hemoglobin derived from erythrocytes within the dissociated tumor. Alternatively, the signal could arise from the reaction of nitric oxide with intracellular heme containing molecules in either host macrophages or tumor cells. The role of erythrocyte contamination in producing the heme-nitrosyl signal was excluded by subjecting freshly dissociated tumor to ammonium chloride erythrocyte lysis (32). This procedure did not diminish the heme-nitrosyl EPR signal which has a lower limit of detectability of approximately 2 p~. Direct measurements of hemoglobin in tumor samples using a colorimetric microtiter assay established that hemoglobin contamination of tumor cell suspensions was undetectable at the 1 pg hemoglobidwell (310 m) sensitivity of the assay (data not shown). These studies showed that hemoglobin was not the source of the heme-nitrosyl EPR signal.
Comparison of EPR Signals from lhmors and from lhmor Znfiltrating Macrophages-The next evaluations were designed to evaluate the relative contributions of host macrophages and tumor cells to the EPR signals. Macrophages grown in culture did not contain EPR active complexes in the absence of cytokines. Macrophage cultures which were activated by IFNy and either lipopolysaccharide or TNFa were found to express only the Fe(RS)2(N0)2 signal as previously reported (17,18). To determine whether either Fe(RS)2(N0)2 or heme-NO signals could be induced in tumor cells in vitro, RD-995 cells grown in vitro were cocultivated with activated macrophages. Activated macrophages were thoroughly washed to remove cytokines before addition to the RD-995 cultures. Cultures containing both RD-995 cells and activated macrophages developed heme-NO and Fe(RS)2(N0)2 EPR signals while those cultured alone or cocultivated with non-activated macrophages did not develop EPR signals. Activated macrophages cultivated alone developed a n Fe(RS)2(N0)2 EPR signal but not a heme-nitrosyl signal. It was not possible to determine whether the Fe(RS)2(N0)2 signal seen in the cocultivated RD-995 and activated macrophage cells developed in the macrophages or in the tumor cells. The presence of MLA, a competitive inhibitor of nitric oxide synthase (1, 2), in the cocultivation medium, diminished formation of both EPR active complexes. DISCUSSION Our current study demonstrates that the cell-mediated immune response is activated during progressive cancer growth in mice. We have also demonstrated that nitric oxide, one of the products of this response, interacts in a complex fashion with iron-containing molecules within the tumor cells. Our EPR data indicates that nitrosylation of iron-dithiol groups to form Fe(RS)2(N0)2 complexes can take place in both tumor cells and macrophages. Two other EPR signals detected in tumor cells, related to the interaction of heme-bound iron with nitric oxide. These signals included a triplet signal centered a t g = 2.012, with a hyperfine coupling constant of 1.7 millitesla which i s typical of heme-nitrosyl complexes in which the iron atom is five coordinate (26,271. A second heme-nitrosyl signal observed in some tumor samples, derived from six coordinate iron in which the hyperfine coupling between iron and NO is not seen (23). The type of EPR signal(s) formed in the tumors did not correlate with the progressor or regressor phenotype of the tumors.
A series of experiments were performed to establish that the heme signals were, in fact, derived from tumor cells. The possibility of hemoglobin contamination was excluded by erythrocyte lysis using Tris-buffered 0.15 M ammonium chloride, and by biochemical determinations of hemoglobin in tumor cell samples. The induction of heme-nitrosyl EPR signals in EMT-6 cells which had been maintained in culture for several years also rules out hemoglobin contamination as the source of this signal. The possibility that these signals were induced in hostderived macrophages was also evaluated. Heme-nitrosyl signals could not be detected in either unstimulated or cytokine-activated macrophages. Addition of cytokine-activated macrophages to cultured tumor cells induced formation of EPR signals in the tumor cells, demonstrating that nitric oxide diffused from the activated macrophages to tumor cell targets, resulting in nitrosylation of iron prosthetic groups. Addition of cytokines to cultured, EPR inactive, tumor cells caused induction of autologous nitric oxide synthesis which directly induced heme-nitrosyl and iron-thiol-nitrosyl EPR signals within the cultured tumor cells. These experiments demonstrate the existence of iron-sulfur and heme-containing targets of nitric oxide in a variety of murine tumors.
The current study also demonstrates that murine cancers elicit an active cell mediated immune response. We have demonstrated both the production of nitric oxide by freshly dissociated tumors and the characteristic EPR signature of NO protein interactions. Nitric oxide synthesis was not induced by the mere proximity of non-activated macrophages with tumor cells. Induction of NO synthesis by cytokine activation is believed to be a two-step process, requiring priming of macrophages (e.g. by IFNy, a lymphocyte product), followed by full cytotoxic activation by inflammatory cytokines such as TNF or IL-1 (31,(38)(39)(40). Our findings imply that other host-derived cells present within murine skin cancers, such as lymphocytes, must be activated by the tumor into a cytokine producing state, and participate in the antitumor response as well. Activated macrophages may also down-regulate lymphocyte mediated immune responses via nitric oxide under some conditions (41,42). The induction of a productive immune response against a cancer may therefore depend on a fine balance between stimulatory and inhibitory signals that act on host cells in the tumor microenvironment. At present it is not known whether cytokine-induced nitric oxide synthesis is associated with stimulation or inhibition of tumor cell proliferation, or whether the EPR signals described here are epiphenomena not associated with either.