Signaling Role of Intracellular Iron in NF-κB Activation*

Iron chelators inhibit endotoxin-induced NF-κB activation in hepatic macrophages (HMs), suggesting a role for the intracellular chelatable pool of iron in NF-κB activation. The present study tested this hypothesis. Analysis of Fe59-loaded HMs stimulated with lipopolysaccharide (LPS), revealed a previously unreported, transient rise in intracellular low molecular weight (LMW)·Fe59complex ([LMW·Fe] i ) at ≤2 min returning to the basal level within 15 min. The [LMW·Fe] i response preceded IκB kinase (IKK) (≥15 min) and NF-κB (≥30 min) activation. Iron chelators (1,2-dimethyl-3-hydroxypyridin-4-one andN,N′-bis-2-hydroxybenzylethylenediamine-N,N′-diacetic acid) abrogated the [LMW·Fe] i response and IKK and NF-κB activation. The [LMW·Fe] i response was also observed in tumor necrosis factor α (TNFα)-stimulated HMs and RAW264.7 cells treated with LPS and interferon-γ but not in primary rat hepatocytes or myofibroblastic cells exposed to LPS or TNFα. Both [LMW·Fe] i response and IKK activation in LPS-stimulated HMs were inhibited by diphenylene iodonium (nonspecific inhibitor for flavin-containing oxidases),l-N 6-(1-iminoethyl)lysine (selective iNOS inhibitor), and adenoviral-mediated expression of a dominant negative mutant of Rac1 or Cu,Zn-superoxide dismutase, suggesting the role of ⋅NO and O 2 ⨪ in mediating the iron signaling. In fact, this inhibition was recapitulated by a cell-permeable scavenger of ONOO−, 5,10,15,20-tetrakis (4-sulfonatophenyl)porphyrinato iron (III) chloride. Conversely, ONOO− alone induced both [LMW·Fe] i response and IKK activation. Finally, direct addition of ferrous iron to cultured HMs activated IKK and NF-κB. These results support a novel signaling role for [LMW·Fe] i in IKK activation, which appears to be induced by ONOO−and selectively operative in macrophages.

NF-B is the prototypic transcription factor in eukaryotic cells known to play a pivotal role in transactivation of promoters for genes involved in inflammation, immune responses, and anti-apoptotic mechanisms (Ref. 1, for review). At least two levels of redox regulation of NF-B appear to exist: one in the nucleus and another in the cytoplasm. The former involves direct redox modification of specific cysteine residues in the DNA binding domain of NF-B. In particular, oxidation of cysteine at the position 62 in p50 inhibits DNA binding activity (2). Conversely, reduction of NF-B by thioredoxin and Ref-1 appear to increase its activity (3,4). For the cytosolic regulation, upstream signaling that is yet to be determined, leads to activation of IKK, 1 resulting in phosphorylation of two serine residues (Ser-32 and Ser-36) on IB␣, its polyubiquitination, and degradation by 26 S proteasome. Treatment of the cells with the antioxidant, N-acetyl-L-cysteine or pyrrolidine dithiocarbamate prevents activation of NF-B (5-7), whereas addition of H 2 O 2 or the generation of O 2 . causes activation of NF-B in certain cell types (5,6,8). TNF␣-induced activation of NF-B is abrogated by inhibition of ROS production by the electron transport chain in mitochondria suggesting oxidant stress from this organelle as a signal for this mode of activation (9). Inhibition of NADPH oxidase blocks or attenuates activation of NF-B in monocytic cells (10,11), whereas inhibitors for 5-lipoxygenase reduce both ROS generation and NF-B activation in lymphoid cells (10), indicating the possible cell type-dependent differences in the source of ROS signals for activation of this transcription factor. It is also important to note that intracellular ROS generation may not be necessary for NF-B activation in all cell types. Indeed, in epithelial cell lines, no detectable increase in ROS generation is observed in association with activation of NF-B (12). Moreover, anti-oxidants and metal chelators are often ineffective in inhibiting NF-B activation in these cells (13). Thus, it appears that a tighter association exists between oxidant stress and NF-B activation in monocytic or lymphoid cells. Hepatic macrophages (HMs), i.e. Kupffer cells, represent the largest population of macrophages, accounting for as much as 70 -80% to total body macrophages. In addition to their crucial role in the first line of defense against invading microorganisms or bacterial products via splanchnic circulation, HMs represent the major site for both regulation of inflammatory and immune responses and metabolism of iron. For the former, HMs release a wide array of soluble factors, including cytokines and chemokines, that are mostly encoded by NF-B-responsive genes as well as growth factors, lipid metabolites, and gaseous mediators (O 2 . , H 2 O 2 , and ⅐ NO). For the latter function, hepatic and splenic macrophages are largely responsible for removal of senescent red blood cells, breakdown of hemoglobin, and recycling of iron for synthesis of new heme or non-heme iron proteins. These two disparate functions of inflammation/immunity and iron metabolism appear mutually related based on several lines of evidence. First, iron is known to affect cellmediated immune response (14). Excessive iron load in macrophages reduces their functions, including phagocytosis, respiratory burst, and cytokine expression (15)(16)(17), the effects likely mediated by iron-mediated toxicity (18). On the other hand, at non-toxic concentrations, iron promotes macrophage differentiation (19) and functions, including anti-microbial effects (20) and TNF-mediated cytotoxicity (21). Iron also increases interleukin-1 secretion by macrophages stimulated by interferon-␥ and LPS (22). On the contrary, the treatment with iron chelators inhibits LPS-mediated NF-B activation and expression of TNF␣ and interleukin-6 by cultured peritoneal macrophages (23) and HMs (24). The finding that iron chelators inhibit activation of NF-B in macrophages suggests a role for a chelatable pool of iron in the signal transduction for this molecular event. In testing this hypothesis, the present study disclosed a novel and transient rise in the intracellular level of low molecular weight⅐iron complex ([LMW⅐Fe] i ) in macrophages stimulated with LPS or TNF␣. This response appears specific to macrophages and precedes IKK activation and increased NF-B binding. The [LMW⅐Fe] i response is tightly correlated with NF-B activation and abrogated by inhibition of iNOS and NADPH oxidase or overexpression of SOD. Our results support the notion that ONOO Ϫ is the effector molecule for inducing [LMW⅐Fe] i release. Last, addition of ionic iron to HMs caused activation of IKK, establishing that iron can serve as an independent agonist and as a potential signaling molecule for IKK activation. Cell Preparation-HMs were isolated from male Wistar rats by the Non-parenchymal Liver Cell Core of the Research Center for Alcoholic Liver and Pancreatic Diseases as previously published (24,25). Briefly, the liver was digested in situ by sequential perfusion with Pronase and collagenase, and non-parenchymal liver cells were fractionated by discontinuous gradient ultracentrifugation using arabinogalactan. An HM-enriched fraction was further purified by the adherence method to achieve the final purity exceeding 95%. After 3 days of culture, the cells were treated with LPS (100 or 500 ng/ml) or TNF␣ (10 ng/ml) in serum-free DMEM in the presence or absence of various inhibitors as described below for subsequent collection of cell lysate for IKK assay and nuclear protein extraction for NF-B electrophoretic mobility shift assay. Hepatocytes were isolated using the standard collagenase digestion technique (26) by the Cell Culture Core of the USC Research A, cultured hepatic macrophages (HMs) were loaded with Fe 59 , washed, and stimulated with LPS (500 ng/ml) in PBS. Cell lysate collected at different time points was subjected to the size exclusion column (Ͻ5000 kDa), followed by scintillation counting of the filtrate. The [LMW⅐Fe] i data are expressed as the percentage to the basal level following standardization by total cellular Fe 59 radioactivity. Note a distinct and transient rise in [LMW⅐Fe] i at 2 min. B, HMs were similarly stimulated with LPS, and cell lysate was collected for an IKK assay with GST-IB␣ as substrate. Note the induction of IKK activity commences at 15 min. C, nuclear proteins were extracted from LPS-treated HMs for electrophoretic mobility shift assay using a B probe. Note an increase in NF-B binding occurs at 30 min. The data presented in A represent means Ϯ S.D. from six experiments. IKK and NF-B binding assays were performed on samples from at least four separate experiments.

Materials-Cell
Center for Liver Diseases. The cells were cultured in DMEM with 10% FCS and used immediately after overnight culture for the experiments described below. RAW264.7 cells were obtained from the American Tissue Culture Collection and cultured in DMEM with 25 mM glucose and 10% FCS. Rat hepatic myofibroblastic cells were isolated from a rat with biliary liver fibrosis by the method described for HMs and by collecting a fraction from the gradient interface between the medium and 1.035. They became spontaneously immortalized, and subsequently cell cones were established by the limiting dilution method. They are cultured in DMEM with 10% FCS. 5% Triton X-100, and 1 mM phenylmethylsulfonyl fluoride. A low molecular mass fraction (Ͻ5000 Da) was prepared by a centrifugation of the lysate in a size-exclusion column (Millipore, MA) at 8600 ϫ g for 30 min at 4°C. Radioactivity of ultrafiltrate or total lysate was determined by a liquid scintillation counter.
Measurement of Intracellular Chelatable Iron in HMs-The chelatable pool of iron in HMs was measured as previously described (27). Briefly, primary cultures of rat HMs were washed with PBS twice, scraped gently, and incubated in PBS (5 ϫ 10 6 /3 ml) to which an iron chelator, desferrioxamine or 1,10-phenanthroline, was added to form a stable complex with iron. The formed complexes were spectrophotometrically detected by their absorption at 430 and 510 nm, respectively. The kinetic of complex formation was assessed with a Beckman spectrophotometer equipped with a flow cuvette connected with a vessel with constant stirring. The blank absorption was obtained by using the mixture of all reagents except chelators.
IKK Assay-To assay the activity of IKK, HMs cultured in a 100-mm dish were treated with LPS (500 ng/ml) for 15-30 min in the presence and absence of the iron chelators, L-NIL, or FeTTPS, washed once with PBS, and lysed with a lysis buffer (20 mM Tris-HCl, pH 7.5, 20 mM NaF, 20 mM ␤-glycerophosphate, 0.5 mM Na 3 VO 4 , 2.5 mM metabisulfite, 5 mM benzamidine, 1 mM EDTA, 0.5 mM EGTA, 10% glycerol and protease inhibitors, 300 mM NaCl, and 1.5% Triton X-100). The lysates were immediately frozen in liquid nitrogen and stored at Ϫ80°C until assayed. IKK activity was determined as previously described (28). HMs were also treated with SNAP, ONOO Ϫ , and FeSO 4 at the indicated concentrations to test their direct effects on IKK. Briefly, IKK was immunoprecipitated by IKK␣ antibodies and protein G-Sepharose. The assay was performed at 30°C for 1 h in buffer containing 20 mM Tris-HCl, pH 7.5, 20 mM MgCl 2 , 2 mM dithiothreitol, 20 M ATP, 2 g of GST-IB␣, and [␥-32 P]ATP (0.5 Ci). The reaction was stopped by addition of Laemmli buffer and was resolved by 10% SDS-PAGE followed by a transfer onto a nitrocellulose membrane. Phosphate incor- To confirm the intracellular release of iron in the intact HMs without detergent lysis of the cells, the cells were stimulated with LPS, and the formation of complexes between iron and added chelators (desferrioxamine (top) and 1-10 phenanthroline (bottom)) was monitored spectrophotometrically in cell suspension. Note the similar transient rise in the formation of the iron⅐chelator complex using either chelator. The peak is slightly shifted right between 5 and 7 min as compared with the response detected with the Fe 59 method. The data represent means Ϯ S.D. from three separate experiments. porated into GST-IB␣ was visualized by analyzing the membrane with a PhosphorImager (Amersham Biosciences).
Immunoblot Analysis-For immunoblot analysis of IKK and nitrated IB␣, cytosolic extracts were prepared as previously reported (29). Proteins were resolved on a 10% SDS-polyacrylamide gel (SDS-PAGE), transferred to a nitrocellulose membrane, incubated with anti-IKK␣ antibodies, and detected by chemiluminescence. For immunoblot analysis of nitrated IB␣, IB␣ was first immunoprecipitated with anti-IB␣ antibodies (Santa Cruz Biotechnology, Santa Cruz, CA), and the immune complex was resolved on a 10% SDS-PAGE and transferred to a membrane for immunoblot analysis using anti-nitrotyrosine antibodies (Upstate Biotechnology, Lake Placid, NY).
Nuclear Protein Extraction and Electrophoretic Mobility Shift Assay-To examine DNA binding by NF-B, nuclear proteins were extracted using the method of Schreiber et al. (30) from cultured HMs, RAW264.7 cells, hepatic myofibroblastic cells, or hepatocytes exposed to LPS or TNF␣ in the presence or absence of the iron chelators. The extracts (5 g) were incubated in a reaction mixture (20 mM HEPES, pH 7.6, 100 mM KCl, 0.2 mM EDTA, 2 mM dithiothreitol, 20% glycerol, 200 g/ml poly(dI-dC)) on ice with the double-strand B consensus sequence (31) or the B site from TNF␣ promoter (32) labeled with 32 P. After a 20-min incubation, the reaction mixture was resolved on a 6% nondenaturing polyacrylamide gel, and the gel was dried for subsequent autoradiography. For the supershift assays, antibodies against p50 and p65 (Santa Cruz Biotechnology) were added to the reaction mixture for an additional 30 min.
Analysis of Nramp1 G169 -expressing RAW264.7-A recent study (33) demonstrates that RAW264.7 cells stably transfected with a vector expressing a wild type Nramp1 G169 (a transformant designated as R37) have lower cellular iron load and a reduced chelatable iron pool as compared with those stably transfected with a vector encoding the gene in an antisense orientation (designated as R21). We have obtained these two stable transformants from Dr. C. H. Barton's laboratory (University of Southampton, UK) and tested whether the altered chelatable pool of iron in R37 affects the [LMW⅐Fe] i signaling, IKK activation, NF-B binding, and TNF␣ release following LPS stimulation. TNF␣ immunoassay was performed with a commercially available mouse TNF␣ enzyme-linked immunosorbent assay kit (R&D Systems, Minneapolis, MN). For the measurement of total non-heme iron content, R37 or R21 cells (3 ϫ 10 7 ) were washed with ice-cold PBS twice. The cells were treated with 1 ml of solution of 10% (w/v) trichloroacetic acid dissolved in 3% (v/v) thioglycolic acid and 1.94 M HCl, vortexed, and put on ice for 10 min. After centrifugation at 14,000 ϫ g for 5 min at 4°C, 100 l of supernatant was incubated for 30 min at 37°C with the equal volume of 0.42 mM bathophenanthroline sulfonate in 2 M sodium acetate. The reaction mixture was then measured for the absorbance at 550 nm on a microtiter plate scanning spectrophotometer (Power Wave 200 TM , Bio-Tek Instruments, Winooski, VT). An iron atomic absorption standard solution (Aldrich) was used as iron standard. The labeling with Fe 59 was performed as described for HMs, the specific activity of Fe 59 in the cells was determined by dividing the radioactivity by iron content, and the chelatable pool of iron was determined by measuring the mobilization of Fe 59 radioactivity into the low molecular weight fraction following the treatment of the cell lysate with L1 (100 M) for 5 min at room temperature.

LPS-induced [LMW⅐Fe]
i -We previously showed that the treatment of cultured HMs with an iron chelator, L1, prevents LPS-induced NF-B activation (24), suggesting the role of the chelatable pool of iron in activation of this transcription factor. To examine this possibility, HMs were loaded with Fe 59 , washed stringently, and stimulated with LPS to determine time-dependent changes in the Fe 59 radioactivity in the low molecular weight cytosolic fraction ([LMW⅐Fe] i ). Addition of LPS (500 ng/ml) resulted in a transient rise in [LMW⅐Fe] i at Յ2 min, followed by a decline to its basal level by 10 -15 min (Fig.  1A). The radioactivity of [LMW⅐Fe] i , separated by a size exclusion column with the molecular mass cutoff of 5000 Da, was 1ϳ2% of total cell-associated radioactivity and increased 2-fold by LPS stimulation. Because the earliest time point we were able to process the cells for this analysis was 2 min, this response might have taken place earlier than 2 min. Kinetic analysis of this event in relation to IKK and NF-B activation revealed that the [LMW⅐Fe] i response preceded the beginning of increased IKK activity at 15 min and that of NF-B binding at 30 min (Fig. 1, B and C). Addition of an iron chelators, L1 (100 M, Fig. 2A) or HBED (100 M, data not shown), prior to LPS stimulation completely abrogated the response as well as IKK (Fig. 2B) and NF-B (Fig. 2C) activation. Because the assessment of a cytosolic labile pool of iron after detergent lysis of the cells may cause the release of iron from solubilized vesicles, organelles, or proteins, we have also employed another approach to determine changes in the chelatable pool of iron in intact cells. This method involved chelation of iron with desferrioxamine or 1,10-phenanthroline and spectrophotometric determination of the iron⅐chelator complexes in cell suspension. Using this method, we confirmed a transient rise in the chelatable iron level with either chelator following LPS stimulation (Fig. 3). However, the response curve shifted to right with a peak occurring between 5 and 7 min, instead of 2 min or earlier when detected by the Fe 59 method. This delay probably reflected the fact that the chelators were not reaching liberated iron immediately and that this spectrophotometric method has a higher detection threshold than the radioactive method. Nevertheless, the early, transient iron response in LPS-stimulated HMs was confirmed by this method in the intact cells.
[LMW⅐Fe] i Response Is Unique to Macrophages-We next tested whether TNF␣, another agonist for NF-B activation, causes the [LMW⅐Fe] i response in HMs. After addition of TNF␣, the identical, transient response to that seen after LPS treatment was reproduced at 2 min (data not shown). We then tested whether the [LMW⅐Fe] i response is observed in other cell types when NF-B is activated. The [LMW⅐Fe] i was examined in cultured hepatocytes treated with LPS or TNF␣. Despite the fact that these treatments caused robust NF-B activation, no [LMW⅐Fe] i response was observed (data not shown). Next, we examined rat liver myofibroblastic cells stimulated with TNF␣ and RAW264.7 cells, the murine macrophage cell line, treated with interferon ␥ (10 units/ml) and LPS (100 ng/ml). Again, no significant changes occurred in the [LMW⅐Fe] i in myofibroblastic cells, although NF-B binding was increased (data not shown). On the other hand, RAW264.7 cells showed a conspicuous [LMW⅐Fe] i peak at 2 min after the treatment preceding activation of NF-B (data not shown). Therefore, these results suggest that the [LMW⅐Fe] i response is unique to macrophages.

Inhibition of iNOS and NADPH Oxidase Abrogates Inhibit [LMW⅐Fe] i Rise-To further examine the association between the [LMW⅐Fe]
i response and NF-B activation and to explore potential mechanisms underlying induction of the [LMW⅐Fe] i response, we tested whether the known inhibitors of NF-B activation could regulate the [LMW⅐Fe] i increase. Cultured HMs were stimulated with LPS in the presence of N-acetylcysteine (NAC, 500 M), diphenylene iodonium (DPI, 1 M), a nonspecific inhibitor for flavin-containing oxidases such as NADPH oxidase and iNOS, and L-NIL (20 M), a selective iNOS inhibitor. DPI and L-NIL effectively blocked the [LMW⅐Fe] i rise, whereas NAC or Me 2 SO as a vehicle had no effect (Fig.  4A). DPI was shown to inhibit NF-B activation in HMs (11), whereas NAC did not inhibit activation and nuclear translocation of NF-B but suppressed NF-B-mediated transcription in HMs (34). Thus, our results for [LMW⅐Fe] i correlate with these findings. L-NIL was shown to attenuate LPS-induced NF-B binding in RAW264.7 cells (35), and our study demonstrates suppression by L-NIL of LPS-induced IKK activity in HMs (Fig.  4B). Thus, these results further tighten the relationship between the [LMW⅐Fe] i response and activation of NF-B, supporting the signaling role of [LMW⅐Fe] i . They also suggest the involvement of iNOS and NADPH oxidase in facilitating this signaling event. Because DPI is not a specific inhibitor, to further test the role of NADPH oxidase, we next used an adenoviral vector expressing a dominant negative mutant of Rac1 (Ad.DN-Rac1) and a vector expressing ␤-galactosidase (Ad.LacZ) as a control. Ad.DN-Rac1 has recently been used to block activation of NADPH oxidase and subsequent activation of NF-B in LPS-stimulated macrophages via its dominant negative effect on the recruitment of the regulatory units of the oxidase to the membrane (36). HMs were infected with Ad.DN-Rac1 followed by stimulation with LPS, and the [LMW⅐Fe] i rise was measured. As shown in Fig. 4C (Fig. 5A). In addition, SNAP did not activate IKK (Fig. 5B) (Fig. 6A) and IKK activation (Fig. 6B). Finally, HMs were directly treated with ONOO Ϫ . Peroxynitrite at the concentrations of 5-50 M activated IKK as early as 5 min (Fig. 6C). This induction of IKK activity is earlier than the timing observed for LPS stimulation, suggesting that ONOO Ϫ is indeed an intermediate signal- ing molecule. More importantly, ONOO Ϫ induced [LMW⅐Fe] i rise (Fig. 6D). LPS and Peroxynitrite Increase Tyrosine Nitration of IB␣-Peroxynitrite is recently shown to activate NF-B by the mechanism that appears to involve nitration of IB␣ (29). If this nitration occurs at Tyr-42, this may abrogate negative regulation of IB␣ degradation facilitated by phosphorylation of this tyrosine residue (37), resulting in accelerated IB␣ degradation. Alternatively, any nitrated IB␣ may be subjected to its preferential degradation (38). Thus, we then examined whether the treatment of cultured HMs with LPS or ONOO Ϫ results in nitration of IB␣. For this analysis, HM cytosolic extracts were immunoprecipitated with an antibody against IB␣ followed by immunoblotting with an anti-nitrotyrosine antibody. As shown in Fig. 6E, LPS induced nitration of IB␣ at 5 and 15 min, and this effect was abrogated when the cells were pretreated with FeTTPS, a decomposition catalyst for ONOO Ϫ or L1, an iron chelator. The treatment of the cells with ONOO Ϫ (10 M) also increased nitration of IB␣ from 15 to 60 min (Fig. 6F). Thus these results support the previous findings on the role of IB␣ nitration as a signaling event for IB␣ degradation besides IKK-mediated phosphorylation of this inhibitory protein (29) and suggest that ONOO Ϫ and iron are involved in this signaling process in the LPS-stimulated cells.
Nramp1 Overexpression Modulates [LMW⅐Fe] i Signaling-Nramp1 is a 90-to 100-kD membrane-spanning protein that is known to confer the natural resistance of macrophages to infection by intracellular pathogens (39) through mechanisms that are believed to regulate iron transport into or out of late endosomes (Ref. 40, for review). A recent study (33) using a stable transformant of RAW264.7 cells expressing high level of Nramp1 revealed that these cells (R37) have less total iron content and a reduced chelatable pool of iron as compared with the mock transfected cells (R21). We considered that the use of these cells would offer a unique opportunity to assess the effects of the reduced chelatable iron pool on LPS-stimulated [LMW⅐Fe] i response and IKK activation. The total non-heme iron content in R37 cells was indeed reduced by 25% (Fig. 7A,  left panel). Following Fe 59 loading, the specific activity achieved with the isotope was comparable between the two cell lines (Fig. 7A, middle panel). The chelatable pool of iron as assessed by an increase in the radioactivity in the low molecular weight fraction after L1 treatment of the cell lysate was significantly reduced in R37 cells (Fig. 7A, right panel), confirming the finding by Atkinson and Barton (33). When these cells were treated with LPS, R37 cells showed attenuated [LMW⅐Fe] i response as compared with R21 cells (Fig. 7B), and this alteration was associated with inhibited IKK activity (Fig.  7C), NF-B binding (Fig. 7D), and TNF␣ release by R37 (Fig.   7E). Thus, these results provide an example of overexpression of a protein that regulates a chelatable pool of iron and in turn modifies iron-mediated signaling, NF-B activation, and expression of an NF-B-responsive gene.
Iron Directly Activates IKK-To test the cause and effect relationship, we next examined the effect of direct addition of ionic iron to cultured HMs on NF-B activation. Ferrous sulfate was added to HMs in the serum-free medium. Addition of ferrous sulfate at the concentration above 10 M induced IKK activity (Fig. 8A), and this effect was apparent as early as 5 min (Fig. 8B). These results demonstrate that iron can serve as an independent agonist for IKK activation. DISCUSSION The present study demonstrated a novel transient rise in the intracellular labile level of iron ([LMW⅐Fe] i ) that leads to IKK and NF-B activation in response to LPS or TNF␣ in macrophages. This response occurs rapidly (Յ2 min) and is transient in nature, much like the intracellular calcium response seen in agonist-induced cells that initiates calcium-mediated signal transduction. Our results suggest that both iNOS-derived ⅐ NO  We have previously reported an opposite example of an enhanced chelatable pool of iron in HMs. In these HMs, which were isolated from a rat model of alcoholic liver disease, the expanded chelatable pool of iron was tightly associated with increased NF-B binding and TNF␣ expression, the changes of which could be entirely normalized by the treatment with the iron chelator, L1 (25). In fact, these cells exhibit a heightened rise in the labile pool of iron after LPS stimulation as compared with the cells from control animals. 2 Thus, these findings collectively demonstrate that the chelatable iron pool within the cells can modulate the [LMW⅐Fe] i signaling for IKK and NF-B activation and that this mode of regulation has physiological and pathological relevance. The source of [LMW⅐Fe] i remains unknown. Three interesting possibilities are suggested. First, ferritin, the main intracellular site for sequestration and storage of excess iron, may be a target for oxygen (42,43) and nitrogen (44) radicals for release of iron. The ONOO Ϫ likely achieves the same effect. Second, iron-containing proteins can be an important source. Even though ⅐ NO is known to mobilize iron from proteins containing the [Fe-S] cluster such as IRP-1 (45), a recent study demonstrates that ONOO Ϫ is more efficient than ⅐ NO in this effect (46). Lastly, iron may be mobilized by ⅐ NO or ONOO Ϫ from the transitional pool prior to incorporation into ferritin (47). Regardless of the source, the [LMW⅐Fe] i response has to be early and transient to serve as the signaling event, because a gradual and sustained increase in [LMW⅐Fe] i seen after the treatment with a ⅐ NO donor, SNAP, did not activate IKK (Fig.  5, A and B). This is very analogous to the differential effects rendered by a transient versus sustained increase in the intracellular calcium concentration (48).
An equally important question is how the transient rise in [LMW⅐Fe] i leads to IKK activation. One possibility is ironmediated release of calcium. Oxidant stress is known to induce calcium mobilization in a variety of cell types (49 -51). This mode of calcium mobilization is prevented with cell-membrane permeable iron chelators (52), suggesting the involvement of iron in this process. In fact, excessive iron loading is known to perturb mitochondrial calcium homeostasis and to induce the release of calcium from this storage site (53). More recently, micromolar concentrations of ferrous iron were shown to cause a transient release of calcium from liver mitochondria without either damaging the organelle or decreasing its membrane potential, suggesting the physiological implication of iron-mediated calcium mobilization (54). Moreover, ONOO Ϫ at a low, physiologically relevant concentration (20 M), induced rapid intracellular mobilization of calcium in thymocytes (55). Because the current study demonstrated the release of iron by ONOO Ϫ (Fig. 6D), an increase in the intracellular calcium concentration may be an immediate consequence initiating the known calcium-mediated signal transduction leading to activation of IKK. Conversely, calcium mobilization may occur upstream of iron. Currently, these possibilities are being investigated. The analogous kinetic patterns of both iron and calcium mobilization may also suggest that iron may signal via the mechanism analogous to that shown for calcium. The transiently released calcium binds to acceptor proteins such as calmodulin, and they in turn recruit and activate signaling proteins. This mode of iron-mediated signaling has never been demonstrated but cannot be ruled out.
Peroxynitrite was recently shown to activate NF-B in monocytes in a manner that appears to involve tyrosine nitration of IB␣ (29), possibly preventing negative regulation mediated via phosphorylation of the tyrosine residues such as Tyr-42 (37). Furthermore, nitrated I B␣ becomes a target for degradation by intracellular enzymes (38). Thus, this nitration-based mode of NF-B activation does not require activation of IKK and phosphorylation of Ser-32 and Ser-36 of IB␣. In fact, our own analysis demonstrates increased nitration of IB␣ by the treatment with LPS or ONOO Ϫ (Fig. 6, E and F), supporting this notion. Furthermore, the nitration induced by LPS is prevented by FeTTPS or an iron chelator, L1, suggesting the involvement of ONOO Ϫ and iron in this signaling event. It has yet to be determined how nitration and phosphorylation IB␣ interact to facilitate NF-B activation in LPS-stimulated HMs and which sites of IB␣ are critical regulatory targets of nitration.
It is interesting that the [LMW⅐Fe] i response was evident 2 S. Xiong, H. She, and H. Tsukamoto, unpublished data. only in HMs and the murine macrophage cell line but not in hepatocytes and myofibroblastic cells, despite the finding that the latter cells also activated NF-B in response to agonists. These results suggest that the iron-mediated signaling may be unique to the macrophage or similar cell type. This notion may not be so out of line. After all, the macrophages are the major site of both iron metabolism and the initiation of inflammatory and immune responses. They specifically utilize iron in their anti-microbial defense by releasing iron extracellularly and promoting Fenton pathway-mediated killing of microbes (20). They also express Nramp1, which regulates efflux of iron from or influx of this metal into late endosomes to help eliminate intracellular pathogens. We propose that macrophages are also evolved to uniquely develop the mechanism by which iron can be utilized as a signaling molecule to support transcriptional regulation of inflammatory and immune-related genes.