Nitrogen Monoxide (NO) and Glucose

Nitrogen monoxide (NO) affects cellular iron metabolism due to its high affinity for this metal ion. Indeed, NO has been shown to increase the mRNA binding activity of the iron-regulatory protein 1, which is a major regulator of iron homeostasis. Recently, we have shown that NO generators increase59Fe efflux from cells prelabeled with59Fe-transferrin (Wardrop, S. L., Watts, R. N., and Richardson, D. R. (2000) Biochemistry 39, 2748–2758). The mechanism involved in this process remains unknown, and in this investigation we demonstrate that it is potentiated upon adding d-glucose (d-Glc) to the reincubation medium. In d-Glc-free ord-Glc-containing media, 5.6 and 16.5% of cellular59Fe was released, respectively, in the presence ofS-nitrosoglutathione. This difference in 59Fe release was observed with a variety of NO generators and cell types and was not due to a change in cell viability. Kinetic studies showed thatd-Glc had no effect on the rate of NO production by NO generators. Moreover, only the metabolizable monosaccharidesd-Glc and d-mannose could stimulate NO-mediated59Fe mobilization, whereas other sugars not easily metabolized by fibroblasts had no effect. Hence, metabolism of the monosaccharides was essential to increase NO-mediated 59Fe release. Incubation of cells with the citric acid cycle intermediates, citrate and pyruvate, did not enhance NO-mediated 59Fe release. Significantly, preincubation with the GSH-depleting agents,l-buthionine-[S,R]-sulfoximine or diethyl maleate, prevented NO-mediated 59Fe mobilization. This effect was reversed by incubating cells withN-acetyl-l-cysteine that reconstitutes GSH. These results indicate that GSH levels are essential for NO-mediated59Fe efflux. Hence, d-Glc metabolism via the hexose monophosphate shunt resulting in the generation of GSH may be essential for NO-mediated 59Fe release. These results have important implications for intracellular signaling by NO and also NO-mediated cytotoxicity of activated macrophages that is due, in part, to iron release from tumor target cells.

Virtually every field of physiology has been influenced by nitrogen monoxide (NO), 1 a small, relatively unstable, potentially toxic, diatomic free radical gas, that is produced in a diverse variety of mammalian cells (for reviews see Refs. [1][2][3]. NO has a physiological role as a short lived messenger molecule and has two principal functions in cells, servoregulation and cytotoxicity. Considering servoregulation, NO is produced in small amounts under physiological conditions and mediates vasorelaxation, regulates blood pressure, controls the adhesion and aggregation of platelets and neutrophils, and is involved in neurotransmission. Most of these actions are mediated through the binding of NO to iron in the heme prosthetic group of soluble guanylate cyclase (1,3). Indeed, the high affinity of NO for iron and other metal ions is a well known branch of coordination chemistry (2).
The importance of iron in mediating the functions of NO is also apparent when examining its cytotoxic effects (3). The cytotoxic functions of NO are observed when it is produced in much larger amounts by macrophages, hepatocytes, and other cells following their exposure to cytokines or microbial products. Interestingly, NO produced via such high output systems inhibits the proliferation of intracellular pathogens and tumor cells. These effects can be explained by the reactivity of NO with iron in the Fe-S centers of important macromolecules, including aconitase and complexes I and II of the electron transport chain (4 -6). The high affinity of NO for iron probably results in both the removal of iron from [Fe-S] centers and the formation of dinitrosyl iron species within [Fe-S] proteins (for review see Ref. 7). NO has been shown to form complexes with a variety of important iron-containing proteins such as ferritin (8), ribonucleotide reductase (9), heme-containing proteins (10 -12), and ferrochelatase (13). In fact, ferritin has been suggested to act as a store of NO (8), and NO-mediated iron release from ferritin has been shown in vitro (14).
Interestingly, co-cultivation of tumor cells with activated macrophages results in the inhibition of target cell DNA synthesis and a concomitant loss of a large fraction (64% per 24 h) of intracellular iron (15). It was speculated that the loss of iron may be due to the NO-mediated release of iron from enzymes such as mitochondrial aconitase (4,16). In contrast to the work of Hibbs and others (4,17,18), it has been suggested that NO targets loosely bound pools of non-heme iron, rather than mitochondrial Fe-S clusters (19). Nonetheless, the relationship between NO and iron is clearly demonstrated by the identifi-cation of Fe-nitrosyl complexes in activated macrophages and their tumor cell targets (17,18,20,21). In these investigations, electron paramagnetic resonance spectroscopy detected signals typical of Fe-dithiol dinitrosyl (Fe(RS) 2 (NO) 2 ) complexes and heme-nitrosyl complexes (17)(18)(19)(20)(21)(22)(23).
The important relationship between NO and cellular iron metabolism is underlined by the fact that NO can also increase the RNA binding of iron-regulatory protein 1 (IRP1), which plays an important role in regulating intracellular iron homeostasis (for reviews see Refs. 3 and 24). The effect of NO on IRP1-RNA binding activity occurs via two possible mechanisms, a direct effect on the [4Fe-4S] cluster and iron mobilization from cells (25)(26)(27)(28)(29). In terms of the ability of NO to increase iron release, we have shown that the NO produced by S-nitroso-N-acetylpenicillamine (SNAP), S-nitrosoglutathione (GSNO), and spermine NONOate (SperNO) can mobilize 59 Fe from prelabeled cells as effectively or more effectively than the clinically used iron chelator desferrioxamine (DFO) (29). However, the mechanism of NO-mediated iron efflux remains unknown (29). It has been hypothesized that NO could be released from cells as a complex composed of NO, iron, and thiol-containing ligands such as cysteine or reduced GSH (23,30,31).
To investigate the mechanism of NO-mediated iron release, we examined the energy dependence of this process. Surprisingly, we found that the metabolizable monosaccharides, Dglucose (D-Glc) and D-mannose (D-Man), can potentiate NOmediated iron efflux from a variety of cell types. In contrast, monosaccharides that cannot be transported into cells or that are poorly metabolized (e.g. L-glucose and D-2-deoxyglucose) have no significant influence. The effect of D-Glc on potentiating NO-mediated 59 Fe release is not due to any change in cellular viability or an effect of D-Glc on NO production by the NO generators. However, these data indicate that the levels of GSH are crucial in terms of NO-mediated iron release. These results suggest that the metabolism of D-Glc by the hexose monophosphate shunt (HMPS) and the maintenance of GSH levels are essential for NO-mediated iron mobilization. Thus, this study demonstrates that there is an important link between energy and iron metabolism and the interaction of NO.

Cell Treatments and Reagents
The NO generator SNAP was synthesized by established techniques (32) from the precursor compound N-acetylpenicillamine (Sigma). L-Buthionine-[SR]-sulfoximine (BSO), cytochalasin B, dipyridyl, D-arabinose, D-allose, D-glucose, D-2-deoxyglucose, D-fructose, D-galactose, Dmannose, D-psicose, D-ribose, D-tagatose, D-xylose, L-glucose, GSH, GSNO, sucrose, and transferrin (Tf) were obtained from Sigma. SperNO was obtained from Molecular Probes (Eugene, OR). Diethyl maleate (DEM) was obtained from Fluka. Minimum essential medium and RPMI 1640 with and without D-Glc were obtained from Life Technologies, Inc. Desferrioxamine (DFO) was obtained from Novartis Pharmaceutical Co., Basel, Switzerland. Pyridoxal isonicotinoyl hydrazone (PIH) and its analogue 2-hydroxy-1-naphthylaldehyde isonicotinoyl hydrazone (311) were synthesized by standard techniques (33). All other chemicals were of analytical reagent quality. The NO generators and other compounds were dissolved in media immediately prior to an experiment. In studies lasting more than 4 h with a NO generator, the media were replaced with fresh solutions to maintain NO levels (29,34).

Cell Culture
The mouse LMTK Ϫ fibroblast cell line was obtained from the European Collection of Cell Cultures (Salisbury, Wiltshire, UK). The human SK-N-MC neuroepithelioma cell line and murine J774 macrophage lines were obtained from the American Type Culture Collection (Manassas, VA). The human HepG2 hepatoma cell line was kindly provided by Dr. Greg Anderson (Queensland Institute of Medical Research, Brisbane, Australia).
The LMTK Ϫ and SK-N-MC cell lines were grown in Eagle's modified minimum essential medium containing 10% fetal calf serum (CSL Ltd., Melbourne, Australia), 1% (v/v) nonessential amino acids (Life Technol-ogies, Inc.), 1 mM sodium pyruvate (Life Technologies, Inc.), 2 mM L-glutamine, 100 g/ml streptomycin (Life Technologies, Inc.), 100 units/ml penicillin (Life Technologies, Inc.), and 0.28 g/ml fungizone (Bristol-Myers Squibb Co.). The J774 and HepG2 cell lines were maintained in RPMI 1640 containing D-Glc with the same additions as described above for minimum essential medium. Cells were grown in an incubator (Forma Scientific) at 37°C in a humidified atmosphere of 5% CO 2 , 95% air and subcultured as described previously (35). Cellular growth and viability were monitored using phase-contrast microscopy, cell adherence to the culture substratum, and trypan blue staining.
In experiments examining the effects of monosaccharides, disaccharides, or pyruvate on 59 Fe release from cells, RPMI medium free of D-Glc and pyruvate was used, and the sugars and pyruvate were added at the concentration required.

Nitrite Determination
The accumulation of nitrite in cell culture supernatants is commonly used as a relative measure of NO production (25,29,34). Nitrite was assayed using the Griess reagent that gives a characteristic spectral peak at 550 nm (36).

ATP Assay
Cellular ATP levels were measured as described by the Sigma ATP assay kit (catalog number 366) with some modifications. In brief, two 75-cm 2 culture flasks containing confluent monolayers of LMTK Ϫ fibroblasts were used for all measurements. After the appropriate incubation with the metabolic inhibitors (cyanide, azide, or rotenone) and/or D-Glc, the cells were removed by incubation in 1 mM EDTA/Ca-Mg-free phosphate-buffered saline, collected, and then counted using a Neubauer counting chamber. After this, the cells were homogenized in 0.3 ml of phosphate-buffered saline and 0.3 ml of 12% trichloroacetic acid by vigorous pipetting. This solution was then placed on ice for 5 min, and the supernatant was separated from the pellet by centrifugation at 3000 rpm/10 min/4°C. The assay was continued as described by the kit.

Glutathione Assay
GSH was measured using the GSH assay kit from Calbiochem. The only modifications to the protocol were that 10-fold more cells (2.4 ϫ 10 7 cells/assay) were used, and the cells were disrupted using two rounds of freeze-thawing before homogenization. These modifications markedly improved the sensitivity of the technique.

Protein Preparation and Labeling
Apotransferrin was labeled with 59 Fe (Dupont NEN) or 56 Fe to produce Fe 2 -transferrin (Tf) using standard procedures (35). In all studies, fully saturated diferric Tf was used. Unbound 59 Fe or 56 Fe was removed by exhaustive vacuum dialysis against 0.15 M NaCl adjusted to pH 7.4 with 1.4% NaHCO 3 (35).

Efflux Assay of 59 Fe from Cells
Efflux Assay, General Protocol-Standard techniques were used to examine the effect of NO and other agents on the efflux of 59 Fe from prelabeled cells (29,33). Briefly, cells were labeled with 59 Fe-Tf (0.75 M) for 2 h at 37°C in D-Glc-containing RPMI. After this incubation, the cell culture dishes were placed on a tray of ice, the medium aspirated, and the cell monolayer washed four times with ice-cold balanced salt solution. The cells were subsequently preincubated for 30 min at 37°C in the presence or absence of the appropriate sugar, and then a final efflux incubation (2 h/37°C) was performed in the presence or absence of the sugar and/or test reagents. After this incubation, the overlying supernatant (efflux medium) was transferred to ␥-counting tubes. The cells were removed from the Petri dishes after adding 1 ml of balanced salt solution and by using a plastic spatula to detach them. Radioactivity was measured in both the cell pellet and supernatant using a ␥-scintillation counter (LKB Wallace 1282 Compugamma, Finland). In some experiments the efflux medium was passed through a 5-kDa molecular mass exclusion filter (Vivaspin 500, Sartorius AG, Germany) to examine the molecular weight of the 59 Fe released.
Iron Efflux Assay, Effect of Metabolic Inhibitors-Established procedures were used to examine the effect of metabolic inhibitors on 59 Fe mobilization (37,38). Briefly, cells were labeled for 3 h at 37°C with 59 Fe-Tf (0.75 M) and washed as described above. The cells were then preincubated for 30 min at 37°C with rotenone (50 M), cyanide (5 mM), or azide (30 mM). This medium was subsequently removed, and the cells were incubated for 3 h at 37°C with medium containing these agents. After this, the overlying supernatant and cells were collected in sepa-rate tubes as described above. Experiments with metabolic inhibitors were performed either in the presence or absence of D-Glc.
Iron Efflux Assay, Effect of Agents That Deplete Reduced Glutathione-We examined the effect on 59 Fe efflux of two agents that deplete GSH, namely BSO and DEM (39,40). The cells were incubated for 20 h at 37°C with BSO (0.01 mM), washed twice, and then labeled with 59 Fe-Tf (0.75 M) for 2 h at 37°C. The cells were then washed 4 times and reincubated for 2 h at 37°C in the presence and absence of GSNO (0.5 mM). After this, the overlying supernatant and cells were separated and placed in ␥-counting tubes. To reconstitute GSH levels, cells treated with BSO were incubated at 37°C with N-acetylcysteine (NAC; 1 mM/20 h) (40,41), washed twice, and then labeled with 59 Fe-Tf (0.75 M) for 2 h at 37°C. The cells were subsequently washed 4 times and reincubated for 2 h at 37°C with GSNO (0.5 mM) in the presence or absence of D-Glc. The overlying supernatant and cells were separated and transferred to ␥-counting tubes.
In contrast to BSO, a 20-h incubation with DEM (1 mM) was cytotoxic; therefore, cells were incubated with DEM (1 mM) for 1 h at 37°C. By using these conditions, no toxic effects were observed. All other experimental conditions were similar to those described for BSO above.
Experimental data were compared using Student's t test. Results were considered statistically significant when p Ͻ 0.05. 59 Fe-Transferrin-In initial experiments the effect of D-Glc was examined on NO-mediated 59 Fe release from prelabeled LMTK Ϫ fibroblasts. These cells were used as our primary model because the effect of NO on their iron metabolism has been well characterized in our laboratory (29,42). Prelabeling of cells with 59 Fe-Tf for 2 h followed by reincubation with media without D-Glc resulted in the release of 1.5 Ϯ 0.1% (3 determinations) of total cellular 59 Fe, whereas the addition of D-Glc to control medium had no effect on 59 Fe mobilization (Fig. 1). Reincubation of prelabeled cells with GSNO in the absence of D-Glc increased 59 Fe efflux to 5.6 Ϯ 0.2% (3 determinations), and this increased to 16.5 Ϯ 1.4% (3 determinations) when D-Glc was added with GSNO ( Fig. 1). This experiment was repeated seven times with a similar result in each case. Statistical analysis of the combined data from seven experiments demonstrated that the addition of D-Glc to GSNO significantly increased (p Ͻ 0.00001) 59 Fe release compared with GSNO alone. There was no change in cellular viability (Ͼ97%) in the presence or absence of D-Glc and/or the NO generators, as determined by trypan blue exclusion, cell adhesion to the culture plate, and overall cellular morphology. The 59 Fe released from the cells in the presence of NO was found to be less than 5 kDa, as determined by Vivaspin 500 size exclusion filters.

The Effect of Monosaccharides and Disaccharides on NOmediated Iron Mobilization from Cells Prelabeled with
Similar results to those reported above for GSNO were also observed for the NO generators SNAP and SperNO, suggesting that the effect was not due to the nature of the NO generator added to cells. The relevant control compounds for GSNO, SNAP, or SperNO, namely GSH, N-acetylpenicillamine, or spermine, respectively, had no effect on 59 Fe release from cells as shown in our previous study (29). Moreover, addition of D-Glc to the control compounds also had no effect on 59 Fe mobilization.
Addition of D-Man to GSNO also enhanced 59 Fe release from cells over that seen with GSNO alone, whereas D-fructose (D-Fuc) had no effect (Fig. 1). Statistical analysis of the results from seven experiments showed that the addition of D-Man to GSNO significantly increased (p Ͻ 0.0006) 59 Fe release compared with GSNO alone. It is relevant to note that D-Man can be metabolized by fibroblasts (43,44), whereas D-Fuc is poorly metabolized (45). To determine whether the effects observed with D-Glc and D-Man were specific to these monosaccharides, a wide variety of different aldohexoses, aldopentoses, ketohexoses, and disaccharides were examined for their ability to enhance NO-mediated 59 Fe mobilization. However, from the range of sugars examined only D-Glc and D-Man were effective. Other monosaccharides (e.g. L-glucose, D-2-deoxyglucose, D-tagatose, D-allose, D-Glc-6-phosphate, etc.) or disaccharides (lactose and sucrose) that are not effectively metabolized or transported across the cell membrane by fibroblasts (45)(46)(47)(48)(49)(50) had no effect on cellular 59 Fe mobilization (Fig. 1). These results suggested that transport of the monosaccharide into the cell and its subsequent metabolism were essential to enhance NO-mediated 59 Fe mobilization.
To test this latter hypothesis further, cells of the hepatocyte lineage (HepG2 hepatoma) were examined to assess if D-Fuc could enhance NO-mediated 59 Fe release. In contrast to fibroblasts, hepatocytes can readily metabolize D-Fuc (45), and using the HepG2 cell line this monosaccharide acted similarly to D-Glc, increasing NO-mediated 59 Fe release to a level 3-fold greater than that observed in its absence (data not shown). This experiment again indicated that metabolism of the monosaccharide was essential for increasing NO-mediated 59 Fe release. Further studies using fibroblasts examined the effect of cytochalasin B that is a potent inhibitor of D-Glc transport into cells (51,52). In these studies, cytochalasin B (50 M) effectively prevented the D-Glc-enhanced increase in 59 Fe release from fibroblasts in the presence of GSNO (Fig. 2), again suggesting that transport of D-Glc into the cell was an essential component of its stimulatory effect.
The effect of D-Glc to increase NO-mediated 59 Fe release from mouse LMTK Ϫ fibroblasts was also seen in a wide variety of cell types including human HepG2 hepatoma cells, human SK-N-MC neuroblastoma cells, and the mouse J774 macrophage cell line (data not shown). These results suggest that the effect of D-Glc on cellular 59 Fe release is not unique to mouse LMTK Ϫ fibroblasts.
Additional experiments examined the effect of D-Glc concentration on NO-mediated 59 Fe release (Fig. 3A). The mobilization of 59 Fe increased as a function of D-Glc concentration and then plateaued at 5 mM which is the physiological concentration of this monosaccharide in serum (53). A similar doseresponse curve was also found with the metabolizable monosaccharide D-Man. The D-Glc-stimulated increase in NOmediated 59 Fe mobilization from cells was observed over reincubation times of 2-6 h (Fig. 3B). In contrast, there was no significant difference in 59 Fe release from control cells in the presence or absence of D-Glc (Fig. 3B). To understand further the mechanism of the NO-mediated 59 Fe mobilization, the ability of D-Glc to increase NO-mediated 59 Fe release was assessed after incubations with 59 Fe-Tf from 15 min to 24 h (Fig. 3C). After all labeling periods with 59 Fe-Tf, D-Glc added with GSNO stimulated 59 Fe release over that seen with GSNO added alone (Fig. 3C). However, as observed with strong iron chelators (54), the greatest increase in 59 Fe release from cells was seen after the shortest labeling period with 59 Fe-Tf (i.e. 15 min; Fig. 3C). Hence, in the present studies, NO appeared to act like an iron chelator in terms of its ability to bind and release intracellular iron. Furthermore, since a much greater proportion of iron taken up from Tf is in a transit form within cells after shorter rather than longer labeling times (54), it can be suggested that NO is binding iron from a labile pool rather than that stored in ferritin.
Effect of Temperature and Metabolic Inhibitors on NO-mediated Iron Efflux-To characterize further NO-mediated iron efflux, the effect of temperature was investigated. As found previously for a number of iron chelators (38), NO-mediated 59 Fe release was temperature-dependent. After a 3-h label at 37°C with 59 Fe-Tf (0.75 M) followed by washing and a 3-h reincubation at 37°C with SNAP (0.5 mM) or the iron chelator pyridoxal isonicotinoyl hydrazone (PIH; 0.5 mM), these agents increased 59 Fe release from 4 Ϯ 1% in control cells to 24 Ϯ 2 and 42 Ϯ 2% (three determinations), respectively. In contrast at 4°C, SNAP or PIH did not increase 59 Fe release over that observed for control cells, which released 6% of cellular 59 Fe (data not shown).
To determine whether NO-mediated 59 Fe efflux was dependent on metabolic energy, the effects of three metabolic inhibitors (azide, cyanide, and rotenone) were examined on 59 Fe mobilization (Fig. 4, A and B). As a relevant control to ensure ATP depletion, the efflux of the chelator PIH was also examined after incubation with the inhibitors. Previous studies have shown that depletion of intracellular ATP levels using rotenone, cyanide, azide, and other inhibitors results in a marked decrease in PIH-59 Fe efflux which occurs by an energy-depend-ent mechanism (37,38). Furthermore, we also assessed ATP levels in cells after exposure to the metabolic inhibitors in the presence or absence of D-Glc (Fig. 4C). In this investigation, cyanide (5 mM), rotenone (50 M), and azide (30 mM) all decreased PIH-and NO-mediated 59 Fe release from cells in the absence of D-Glc, the most marked effects being observed after incubation with rotenone and azide (Fig. 4A). Statistical analysis of the data from 4 experiments showed that when cells were incubated with metabolic inhibitors in the absence of D-Glc there was a significant (p Ͻ 0.01) decrease in both NOand PIH-mediated 59 Fe release and a significant (p Ͻ 0.0001) decrease in ATP levels when compared with the control.
Addition of D-Glc to cyanide and azide largely prevented their inhibitory effects on NO-and PIH-mediated 59 Fe mobilization (compare Fig. 4, A and B) and reconstitutes cellular ATP levels (Fig. 4C). In contrast, D-Glc does not prevent the effect of  59 Fe release from labeled cells. LMTK Ϫ fibroblasts were labeled as described in Fig. 1, washed, and then reincubated with various concentrations of D-Glc (0.1-11 mM) in the presence or absence of the NO generator GSNO (0.5 mM). B, D-Glc stimulates 59 Fe release in the presence of GSNO but not control medium for reincubation periods of 2-6 h. Cells were labeled as described in Fig. 1, washed, and then reincubated for 0.5-6 h at 37°C in the presence and absence of D-Glc (11 mM) and/or GSNO (0.5 mM). C, D-Glc-stimulated 59 Fe release in the presence of NO is greatest after short labeling periods with 59 Fe-transferrin. Cells were labeled with 59 Fe-transferrin (0.75 M) for 15 min, 3 h, or 24 h at 37°C, washed, and then incubated for 30 min in the presence or absence of D-Glc (11 mM). This medium was removed and the cells then reincubated for 2 h at 37°C in the presence and absence of D-Glc and/or GSNO. After this, the overlying medium (efflux) and cells were collected in separate tubes. Results in A and B or C are means Ϯ S.D. of 3 determinations in a typical experiment of 2 performed. rotenone at inhibiting NO-mediated 59 Fe release (compare Fig.  4, A and B), even though D-Glc prevented the inhibitory effect of rotenone on PIH-mediated iron mobilization (Fig. 4B) and helped to reconstitute cellular ATP levels (Fig. 4C). These data suggest that the mechanism of action of rotenone at inhibiting NO-mediated 59 Fe release is different from that of cyanide and azide. Collectively, these results suggested that like 59 Fe efflux mediated by PIH (37, 38), metabolic energy was necessary for NO-mediated 59 Fe mobilization from cells.
Effect of D-Glucose on the Mobilization of Iron Mediated by Strong Iron Chelators-Considering the observation that NO appeared to act like an iron chelator to increase 59 Fe release (see Ref. 29 and data above), experiments were performed to assess whether the increase in 59 Fe efflux in the presence of D-Glc could also be observed with other strong iron chelators (Fig. 5). Interestingly, D-Glc only stimulated 59 Fe release in the presence of NO, the addition of D-Glc having no effect on 59 Fe release mediated by PIH or dipyridyl (Fig. 5). Similarly, D-Glc had no effect on the ability of DFO (0.5 mM) to increase 59 Fe release from labeled LMTK Ϫ or SK-N-MC neuroblastoma cells (data not shown). Together, these results suggested that the effect of D-Glc on increasing 59 Fe mobilization is relatively specific for NO and was not an effect observed with iron chelators in general.
The Effect of D-Glucose on the Rate of Nitrite Production by NO-generating Agents-It was possible that the stimulatory effect of D-Glc on NO-mediated 59 Fe mobilization from cells was due to the ability of this agent to increase NO release from the NO generators as a function of time. To ensure that this was not the case, the NO generators were incubated with D-Glc in the presence and absence of cells, and the production of nitrite was determined as a function of time (Fig. 6). Nitrite was assayed as it is the direct result of NO oxidation and has been widely used to assay NO production by NO generators (55,56). In each case, the addition of D-Glc to the NO generators had no effect on nitrite production (Fig. 6). Nitrite production by GSNO and SNAP increased linearly as a function of time, whereas nitrite generated by SperNO increased up to 15 min and then plateaued in the absence of cells or slightly decreased in the presence of cells (Fig. 6). The latter slight decrease in nitrite levels after incubation of cells with SperNO was reproducible over 3 separate experiments. Previous studies have shown that the mechanism of NO release from SperNO is different from the other NO generators examined (57).
The Effect of Pyruvate, Citrate, and Reduced Glutathione on NO-mediated Iron Mobilization-D-Glc is metabolized in cells by two main pathways, via the glycolysis/citric acid cycle or by the hexose monophosphate shunt (HMPS). As the D-Glc-mediated increase in NO-mediated iron mobilization from cells could be due to the metabolism of this monosaccharide by either pathway, the effect on iron mobilization of intermediates from both was examined. To assess if the citric acid cycle may be involved, cells were incubated with pyruvate or citrate (1 or 11 mM) (Fig. 7). Both of these latter substrates are metabolized by the tricarboxylic acid cycle and can be efficiently transported into cells and metabolized (58 -62). In contrast to D-Glc that markedly increased 59 Fe release, pyruvate or citrate had no effect (Fig. 7).  (11 mM). After this preincubation, the NO generator GSNO (0.5 mM) or the iron chelator PIH (50 M) was added in the absence (control) or presence of either cyanide (5 mM), rotenone (50 M), or azide (30 mM), and the cells were reincubated for 3 h at 37°C. After this, the overlying medium (efflux) and cells were collected in separate tubes. C, to examine ATP levels, cells incubated as described above in the absence or presence of D-glucose and the metabolic inhibitors cyanide, azide, or rotenone, were harvested and assayed for ATP levels as  59 Fe release from prelabeled cells in the presence of the NO generator GSNO but not strong iron chelators. LMTK Ϫ cells were labeled as described in Fig. 1, washed, and then preincubated for 30 min at 37°C in medium with and without D-Glc (11 mM). After this, the medium was removed and the cells then reincubated for 2 h at 37°C in the presence and absence of D-Glc (11 mM) and GSNO (0.5 mM), PIH (50 M), or dipyridyl (0.5 mM). After this, the overlying medium (efflux) and cells were collected in separate tubes. Results are means Ϯ S.D. of 3 determinations of a typical experiment of 3 performed.
When D-Glc is metabolized by the HMPS, there is an increase in the level of NADPH and subsequently the ratio of GSH over GSSG increases (41). Considering this, the change in intracel-lular redox state may have a critical effect on intracellular iron metabolism. We hypothesized that the metabolism of D-Glc by the HMPS results in an elevation of GSH levels that may affect the redox state of the cell and/or the access of NO to chelatable 59 Fe pools. Indeed, previous studies have shown that NO depletes intracellular GSH and then activates the HMPS (63). Similarly, in the present experiments, incubation with NO decreased cellular GSH levels in the absence of D-Glc, whereas when D-Glc was added to the NO generator it markedly prevented GSH depletion (Fig. 8). Statistical analysis of the results from 3 experiments demonstrated that NO caused a significant decrease (p Ͻ 0.0001) in GSH levels in the absence of D-Glc. Hence, these data suggest that D-Glc acts as a metabolic substrate to largely prevent GSH depletion in the presence of NO.
Considering that the level of GSH could be a critical factor in NO-mediated iron mobilization, experiments were designed to manipulate the levels of intracellular GSH and assess its effect on 59 Fe mobilization. Cells were depleted of GSH using BSO which is a highly selective and potent inhibitor of the enzyme ␥-glutamylcysteine synthetase that is involved in GSH synthesis (39) (Fig. 9, A-C). Preincubation of cells with BSO (0.01 mM) alone had no appreciable effect on 59 Fe release compared with control cells incubated with medium alone, whereas BSO markedly inhibited NO-mediated 59 Fe efflux (Fig. 9A). In the same experiment, to reconstitute cellular GSH levels after exposure to BSO, cells were then incubated with NAC (1 mM) for 20 h, and the effect on NO-mediated 59 Fe release was examined. Treatment of BSO-treated cells with NAC slightly increased 59 Fe release compared with the control, whereas NAC totally reconstituted NO-mediated 59 Fe release (compare Fig.  9, A and B). Measurement of cellular GSH concentrations demonstrated that incubation with BSO reduced GSH levels to less than 40% that seen with the untreated control (Fig. 9C). In addition, when NAC was added to BSO-treated cells, it reconstituted GSH to a comparable level of that observed in cells exposed to NAC alone (Fig. 9C). Statistical analysis of the results from 4 experiments showed that BSO significantly (p Ͻ 0.00001) reduced GSH levels compared with the control, and treatment of BSO-treated cells with NAC significantly (p Ͻ 0.00001) increased GSH.
To confirm the results with BSO, experiments were performed by incubating cells for 1 h at 37°C with the oxidizing agent DEM (1 mM) which is effective at decreasing cellular GSH levels (40,64,65). Treatment with DEM resulted in very similar results to those reported with BSO above, significantly (p Ͻ 0.0001) decreasing GSH levels and inhibiting NO-mediated 59 Fe release in 3 separate experiments (data not shown). Collectively, the results with BSO and DEM indicated that GSH is essential for NO-mediated 59 Fe mobilization from cells.
In good agreement with the results performed with chelators in the presence or absence of D-Glc (see Fig. 5), a 20-h incubation with BSO markedly decreased 59 Fe release after exposure to GSNO, viz. from 18.2 Ϯ 0.5 to 4.6 Ϯ 1.6% (3 determinations) (Fig. 10). In contrast, BSO had no effect on 59 Fe mobilization mediated by DFO (0.5 mM), dipyridyl (0.5 mM), PIH (50 M), or 311 (50 M) (Fig. 10). These results indicate that the involvement of GSH in 59 Fe release was an effect specific for NO rather than iron chelators in general.

DISCUSSION
This is the first study to demonstrate that NO-mediated iron mobilization from cells can be markedly influenced by the metabolism of D-Glc. The results are important because they demonstrate a unique relationship between glucose and iron metabolism. This link has previously been speculated to exist based on the finding that IRP1 acts as an RNA-binding protein in the absence of iron and also has aconitase activity when sufficient iron is present (24).
Our studies demonstrate that the effect of D-Glc at stimulating NO-mediated iron mobilization is due to its ability to be transported into the cell and then subsequently metabolized. Mono-and disaccharides that are not transported though cell membranes or that are poorly metabolized have no effect (Fig.  1). Moreover, the effect of D-Glc at increasing NO-mediated iron release was not due to this monosaccharide acting as an iron chelator, since D-Glc had no effect on iron mobilization from control cells ( Fig. 1 and Fig. 3, A and B). In addition, it is of interest that D-tagatose and D-Fuc can chelate iron (50) but did not increase NO-mediated iron release, in contrast to D-Glc and D-Man (Fig. 1). Again, these results suggest that the metabolism of the monosaccharide is important. The effect of D-Glc at stimulating NO-mediated 59 Fe release appears to be due, at least in part, to the presence of reduced GSH within the cell (Figs. 8 and 9). Previous studies have also shown that in the absence of D-Glc there is a decrease in reduced GSH (41). This latter effect is due to depletion of NADPH derived from the HMPS that is necessary for the conversion of oxidized glutathione (GSSG) to its reduced counterpart (41).
We have clearly demonstrated that incubation of cells with NO decreased cellular GSH levels in the absence of D-Glc (Fig.  8). In contrast, when D-Glc was added with the NO generator it markedly prevented the depletion of GSH (Fig. 8). Our results are strongly supported by previous studies, which showed that NO rapidly caused depletion of intracellular GSH due to the formation of intracellular S-nitrosoglutathione which then results in activation of the HMPS (63). In the presence of NO and the absence of D-Glc, the ability of the HMPS to reconstitute cellular GSH levels via the production of NADPH does not appear efficient (Fig. 8). Our results add significantly to the observations of Clancy and associates (63) and have implications for both intracellular signaling via the NO-Fe interaction, intermediary metabolism, and NO-mediated cytotoxicity. For instance, the fact that NO markedly influences intracellular iron metabolism may be due, in part, to changes in GSH levels. This in turn could have major effects on pathways that rely on iron-containing enzymes that are vital for energy production (e.g. mitochondrial aconitase; Ref. 4) and DNA synthesis (e.g. ribonucleotide reductase; Ref. 9). Indeed, it is well known that exposure of tumor cells to NO results in the inhibition of energy production and DNA synthesis due to the action of NO on these and other iron-containing proteins (4,9,16). This may be important for understanding the large loss of cellular iron observed from tumor target cells (64% after 24 h) after exposure to activated macrophages producing NO (15). For instance, the increase in HMPS activity mediated by NO stimulates GSH synthesis in tumor cells, which then supplies GSH which is involved in iron release.
The process whereby GSH can facilitate NO-mediated 59 Fe mobilization could be due to a number of possible mechanisms. For example, GSH may alter cellular redox state by reducing the levels of oxidants such as hydrogen peroxide (41) which may change the intracellular distribution of iron or lead to an increase in the proportion of Fe(II) that may be preferentially bound by NO. However, this hypothesis does not appear compelling, because in contrast to NO-mediated iron efflux, iron mobilization by the Fe(II) chelator dipyridyl or other chelators was not affected after incubation with either D-Glc or BSO (Figs. 5 and 10). Alternatively, the increase in GSH levels after incubation with D-Glc could result in GSH acting as a ligand that together with NO would complete the coordination shell of iron. Such a "mixed iron complex" with both NO and GSH ligands bound to iron may provide an appropriate lipophilic balance to allow diffusion through the cell membrane or transport by an appropriate carrier. Regarding this, it is of considerable interest that electron paramagnetic resonance spectroscopy can detect signals typical of Fe-dithiol dinitrosyl complexes (Fe(RS) 2 (NO) 2 ) in cells exposed to NO (17)(18)(19)(20)(21)(22)(23).
It is probable that the effect of D-Glc on stimulating NOmediated iron mobilization from cells is not just due to its effect on GSH metabolism. Indeed, our experiments have shown that like the iron chelator PIH, NO-mediated 59 Fe release is both temperature-and energy-dependent, suggesting a membrane transport mechanism could be involved. Whereas the process that may be responsible for NO-mediated iron efflux has yet to be identified, it is of interest that an iron export molecule known as ferroportin 1 has recently been cloned (66). Whether this transporter can export iron bound to NO or PIH remains a question that requires further investigation.
In summary, this study has demonstrated that NO-mediated iron mobilization can be potentiated by D-Glc due to the transport and metabolism of this monosaccharide. This effect was observed with a range of NO generators (GSNO, SperNO, and SNAP) and cell types. Significantly, the increase in NO-mediated iron release is dependent on the presence of reduced GSH within the cell. Our results clearly demonstrate that there is a relationship between D-Glc and iron metabolism. This may have important implications for intracellular signaling via NO and also NO-mediated cytotoxicity of activated macrophages that is due, in part, to iron release from tumor target cells.