The Phase 2 Enzyme Inducers Ethacrynic Acid, DL-Sulforaphane, and Oltipraz Inhibit Lipopolysaccharide-Induced High-Mobility Group Box 1 Secretion by RAW 264.7 Cells

High-mobility group proteins are small DNA binding proteins that serve an important role in transcriptional regulation (Bustin et al., 1990). When present in the extracellular milieu, one of these proteins, high-mobility group box 1 (HMGB1), also functions as a cytokine-like molecule. For example, HMGB1 promotes TNF release from mononuclear cells (Andersson et al., 2001) and induces DNA binding by the proinflammatory transcription factor NF-κB and inducible nitric-oxide synthase expression in Caco-2 human enterocyte-like cells (Sappington et al., 2002). Additionally, HMGB1 promotes adhesion molecule expression on cultured human umbilical venular endothelial cells (Treutiger et al., 2003). HMGB1 is actively secreted by immunostimulated macrophages (Wang et al., 1999; Gardella et al., 2002; Bonaldi et al., 2003; Rendon-Mitchell et al., 2003) and is also released by necrotic but not apoptotic cells (Scaffidi et al., 2002).

Extracellular HMGB1 has been implicated in the pathogenesis of inflammatory diseases in both experimental animals and humans. When mice are injected with lipopolysaccharide (LPS) or rendered septic by cecal ligation and puncture, circulating levels of HMGB1 increase dramatically, albeit only after at least 16 h have elapsed after the onset of endotoxemia or infection (Wang et al., 1999; Yang et al., 2004). High levels of HMGB1 have been detected in synovial fluid specimens from patients with rheumatoid arthritis (Kokkola et al., 2002; Taniguchi et al., 2003) and serum samples from patients with severe sepsis (Wang et al., 1999; Sunden-Cullberg et al., 2005). Treatment of endotoxemic (Wang et al., 1999) or septic (Yang et al., 2004) mice with neutralizing anti-HMGB1 antibodies improves survival even when treatment is started as late as 24 h after the onset of endotoxemia or infection. Similarly, administering neutralizing anti-HMGB1 antibodies to rodents with collagen-induced arthritis ameliorates the clinical and histological features of the disease (Kokkola et al., 2003).

In view of the apparent importance of HMGB1 as a mediator of inflammatory diseases, compounds capable of blocking the secretion of this protein from immunostimulated macrophages might prove to be beneficial therapeutic agents. One such compound is the simple ester ethyl pyruvate (Ulloa et al., 2002). Cholinergic agonists that activate the α7 nicotinic acetylcholine receptor are another class of compounds that have been shown to inhibit HMGB1 release from immunostimulated macrophages (Wang et al., 2004). In the current report, we present data showing that incubating LPS-stimulated RAW 264.7 murine macrophage-like cells with ethacrynic acid (EA) inhibits the release of several proinflammatory mediators, including HMGB1. Because EA is an electrophilic Michael acceptor and inducer of so-called phase 2 enzymes (Ciaccio et al., 1994; Tjalkens et al., 1998), we sought to determine whether other phase 2 enzyme inducers also inhibit LPS-induced HMGB1 secretion. Herein, we also present data showing that two such compounds, namely, oltipraz (Buetler et al., 1995) and dl-sulforaphane (Zhang et al., 1994; Gerhuaser et al., 1997), inhibit HMGB1 release from LPS-stimulated RAW 264.7 cells.

Materials and Methods

Materials. All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise noted. dl-Sulforaphane, U0126, SN50 (amino acid sequence AAVALLPAVLLALLAPVQRKRQKLMP), and SC-514 were obtained from Calbiochem (San Diego, CA). Oltipraz was obtained from National Cancer Institute (Bethesda, MD). The anti-HMGB1 antibody was from BD PharMingen (San Jose, CA).

Cell Culture. The murine macrophage-like RAW 264.7 cell line was obtained from the American Type Culture Collection (Manassas, VA). The cells were maintained in Dulbecco's modified essential medium supplemented with 10% fetal bovine serum (FBS; Hyclone Laboratories, Logan, UT), 100 units/ml penicillin, and 100 μg/ml streptomycin at 37°C in a humidified incubator (5% CO₂, 95% air). For stimulation experiments, cells were transferred to either 50-mm culture dishes or 24-well polystyrene culture plates at 0.3 or 1 × 10⁶ cells/well (depending on the experiment) in 1 ml of medium per well. After overnight incubation, the medium was removed and replaced with Dulbecco's modified essential medium containing 5% FBS (for most experiments) or 0.25% FBS (for experiments design to measure HMGB1 in conditioned media). Cells were stimulated by adding 10 to 1000 ng/ml Escherichia coli LPS (serotype O111:B4) in the presence or absence of graded concentrations of EA, dl-sulforaphane, oltipraz, or pyrrolidine dithiocarbamate (PDTC).

Nitrite Assay. RAW 264.7 cells were incubated for 24 h in the presence or absence of LPS with or without graded concentrations of EA. Nitrite concentrations were measured in cell supernatants as an indicator of nitric oxide production using a commercially available Griess reaction kit (Oxis International, Portland, OR).

TNF Assay. RAW 264.7 cells were preincubated with or without various concentrations of EA for 4 h before stimulation with LPS for 2 h. TNF was measured in cell culture supernatants using a commercially available enzyme-linked immunosorbent assay (ELISA) kit from BD Biosciences (San Diego, CA), according to the manufacturer's specifications.

IL-6 Assay. RAW 264.7 cells were incubated for 24 h in the presence or absence of LPS with or without EA. IL-6 concentrations were measured in cell supernatants using a commercially available ELISA kit from BD Biosciences, according to the manufacturer's specifications.

Western Blot Assay for HMGB1. Equal volumes of cell culture supernatant were mixed with Laemmli buffer (20% glycerol, 10% β-mercaptoethanol, 5% SDS, 0.2 M Tris·HCl, pH 6.8, and 0.4% bromphenol blue). After boiling for 10 min, the samples were subjected to 10% SDS-polyacrylamide gel electrophoresis. The resolved proteins were transferred to Hybond-P polyvinylidene difluoride membrane (Amersham Biosciences UK, Ltd., Little Chalfont, Buckinghamshire, UK), blocked with bovine lacto-transfer optimizer buffer [1× phosphate-buffered saline (PBS), 5% nonfat milk, and 0.05% Tween 20, and in some cases, 0.2% NaN₃] for 1 h. The membrane was then incubated with rabbit polyclonal anti-HMGB1 antibodies (BD PharMingen) 1:2000 diluted in blocking buffer overnight at 4°C. After washing three times in 1× PBST, immunoblots were exposed at room temperature for 1 h to a 1:20,000 dilution of the horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (Sigma-Aldrich). After washing three times with 1× PBST, the membrane was illuminated with the ECL substrate (Amersham Biosciences UK, Ltd.) and X-ray film-exposed, according to the manufacturer's instructions. Densitometry was performed using Gene Tools 3.05 (SynGene, Cambridge, UK).

Western Blot Assay for Total and Phosphorylated ERK1/2. For preparation of samples from the in vitro studies, RAW 264.7 cells were lysed in 1 ml of radioimmunoprecipitation assay buffer, and the extracted proteins removed from the tissue culture plate by gentle scraping with a rubber policeman and transferred to a 1.5-ml microcentrifuge tube. The samples were sonicated three times for 30 s on ice using a 0.1-W Fisher Scientific Sonic dismembrator fitted with a microtip on power setting 3. The lysate was transferred to a microcentrifuge tube and incubated for 30 min on ice. The lysate was centrifuged at 10,000g for 20 min at 4°C, and then the supernatant was transferred to a new tube. Total protein concentration was determined using the Bio-Rad (Hercules, CA) protein reagent.

Equivalent amounts of protein were mixed with Laemmli buffer. After boiling 5 to 10 min, the protein samples were centrifuged for 10 s. Samples of the supernatants containing 100 μg of protein per lane were electrophoresed at 100 mA for 40 min on 7.5% precast SDS-polyacrylamide gels (Bio-Rad). The size-fractionated proteins were electroblotted onto a Hybond-P polyvinylidene difluoride membrane and blocked with bovine lacto-transfer optimizer buffer (1× Tris-buffered saline, 5% milk, 0.05% Tween 20, and 0.2% NaN₃) for 60 min. The membranes were then incubated at 4°C overnight with rabbit anti-phospho-ERK1/2 (1:1000 dilution) or anti-ERK1/2 (1:1000 dilution) polyclonal antibody (Cell Signaling Technology Inc., Beverly, MA). Dilutions of antibody were made using Tris-buffered saline/Tween 20 (1× TBS, 0.1% Tween 20, and 5% bovine serum albumin). After washing three times in 1× PBST, immunoblots were exposed at room temperature for 1 h to a 1:10,000 dilution of horseradish peroxidase-conjugated anti-Ig secondary antibody. After three washes in PBST and two washes in PBS, the membrane was impregnated with the ECL substrate and used to expose X-ray film according to the manufacturer's instructions.

Electrophoretic Mobility Shift Assay. Cells were harvested and centrifuged at 1000g for 5 min. The pellet was resuspended in 400 μl of buffer I (10 mM Tris, pH 7.8, 5 mM MgCl₂, 10 mM KCl, 0.3 mM EGTA, 0.5 mM dithiothreitol, 0.3 M sucrose, 1 mM phenylmethylsulfonyl fluoride, 1.0 mM sodium orthovanadate, and 1× mammalian protease inhibitor cocktail; Sigma-Aldrich catalog no. P8340). The tubes were placed on ice for 15 min to allow the cells to swell and facilitate lysis. Nonidet P-40 was added to the final concentration of 0.5% (v/v), and the tubes were vortexed at full speed for 10 s. Nuclei were harvested by centrifugation at 7200g for 10 s at 4°C. The supernatant was aspirated, and the nuclei were resuspended in 50 μl of buffer II (20 mM Tris, pH 7.8, 5 mM MgCl₂, 320 mM KCl, 0.2 EGTA, 0.5 mM dithiothreitol, and the mixture of protease inhibitors described above). Nuclear proteins were extracted for 15 min on ice followed by centrifugation at 13,500g for 15 min. The protein concentration was determined using the Bio-Rad protein concentration reagent (Bio-Rad).

The EMSA for NF-κB was used to measure DNA binding activity using double-stranded oligonucleotides as published previously (Delude et al., 1994). Briefly, the sequence of the double-stranded NF-κB oligonucleotide was as follows: sense, 5′-AGT TGA GGG GAC TTT CCC AGG C-3′ and antisense, 3′-TCA ACT CCC CTG AAA GGG TCC G-5′ (E3291; Promega, Madison, WI). The oligonucleotides were end-labeled with [γ-³²P]ATP (PerkinElmer Life and Analytical Sciences, Boston, MA) using T4 polynucleotide kinase (Promega). Three micrograms of nuclear protein per reaction was incubated with radiolabeled NF-κB probe in bandshift buffer (10 mM Tris, pH 7.8, 40 mM KCl, and 1 mM EDTA) in the presence of 2 μg of poly(dI-dC) for 20 min at room temperature. The binding reaction mixture was electrophoresed on 4% nondenaturing polyacrylamide gels, which were then dried and used to expose Kodak X-Omat AR film (Eastman Kodak, Rochester, NY) at –80°C overnight using an intensifying screen.

Immunofluorescence Confocal Microscopy. RAW 264.7 cells were grown on 12-mm glass coverslips. After incubation for 24 h in the absence or presence of LPS (in the absence or presence of other compounds), the cells were fixed in 2% paraformaldehyde in PBS for 1 h. The cells were washed three times in PBS and then three times (5 min each) with PBS supplemented with 0.5% bovine serum albumin (PBB). The cells were permeabilized in 0.1% Triton X-100 in PBB for 30 min and then washed once with PBB. Cells were blocked with 2% bovine serum albumin in PBS for 40 min and then washed once with PBB. Rabbit anti-HMGB1 antibodies (1:2000) were added to cells in PBB incubated at room temperature for 1 h. Cells were washed four times in PBB, and then secondary antibodies (goat anti-rabbit Alexa 488; 1:500 dilution; Molecular Probes, Eugene, OR), rhodamine-phalloidin (1:250 dilution; Molecular Probes), and the nuclear dye DRAQ5 (1:1000 dilution; Biostatus, Leicestershire, UK) were added for 1 h at room temperature. The cells were washed three times in PBB and then three times in PBS. The cells were mounted using gelvatol [23 g of poly(vinyl alcohol) 2000, 50 ml glycerol, and 0.1% sodium azide to 100 ml of PBS] and then viewed on a confocal scanning fluorescence microscope (Fluoview 1000; Olympus, Malvern, NY).

Image Analysis. The density of cytoplasmic staining by the anti-HMGB1 antibody was assessed using Scion Image for Windows (Scion Corporation, Frederick, MD). Images depicting only the green (HMGB1) channel were imported into the program as TIFF files. The regions of interest (i.e., cells and cell nuclei) were delineated by hand using a computer mouse. To compute the density of cytoplasmic HMGB1 staining in a given cell, the density of HMGB1 staining in the nucleus was subtracted from the density of HMGB1 staining in the whole cell. This process was carried out for 20 to 22 cells on three or four fields for each condition studied (control, LPS, LPS + EA, and LPS + dl-sulforaphane). Results are reported as mean ± S.E.M. density for each condition.

Statistical Analysis. All Western blot and EMSA experiments were repeated at least three times with nearly identical results, and representative blots and gel shifts are shown in the figures. Data are presented as means ± S.E.M. Changes in variables for different assays were analyzed via one-way analysis of variance with a least significant difference post hoc test. Differences were considered to be statistically significant at P < 0.05.

Results

EA Inhibits LPS-Induced TNF Secretion. RAW 264.7 murine macrophage cells were preincubated for 2 h in the absence or presence of graded concentrations of EA. After being incubated for an additional 2 h in the presence of LPS (10 ng/ml), culture supernatants were harvested and assayed for TNF concentration by ELISA. Control cultures were incubated for 4 h in the absence of either EA or LPS. As expected, incubating RAW 264.7 cells with LPS in the absence of EA promoted TNF secretion (Table 1). EA inhibited LPS-induced TNF secretion in a concentration-dependent manner. Even at the lowest concentration tested (7.5 μM), EA significantly inhibited secretion of the cytokine, and the highest concentration tested (120 μM) almost completely inhibited TNF secretion. EA-induced cell death (as assessed by exclusion of the vital dye trypan blue) was not observed under any of the conditions examined (data not shown).

EA Inhibits LPS-Induced IL-6, IL-10, and Nitric Oxide Secretion. RAW 264.7 cells were treated with 10 ng/ml LPS in the absence or presence of graded concentrations of EA for 24 h. Control cultures were incubated for 24 h in the absence of either EA or LPS. At the end of this period, concentrations of IL-6, IL-10, and nitrite (a marker for nitric oxide production) were measured in the conditioned media. EA inhibited IL-6, IL-10, and nitric oxide release in a concentration-dependent manner (Table 2). Secretion of both IL-10 and nitric oxide was significantly inhibited by EA at the lowest concentration tested (5 μM), whereas the lowest effective concentration for inhibiting IL-6 secretion was 10 μM. EA-induced cell death (as assessed by trypan blue dye exclusion) was not observed under any of the conditions examined (data not shown).

Transient Exposure to EA Inhibits LPS-Induced IL-6, IL-10, and Nitric Oxide Secretion. We sought to determine whether transient exposure of cells to EA leads to persistent inhibition of LPS-induced secretion of pro- or anti-inflammatory mediators. Accordingly, RAW 264.7 cells were pretreated with graded concentrations of EA for 1 h, and then the cells were washed extensively before being exposed for 24 h to 10 ng/ml LPS. At the end of the second period of incubation, concentrations of IL-6, IL-10, and nitrite were measured in the conditioned media. Control cultures were incubated for 24 h in the absence of either EA or LPS. As was the case with persistent exposure to EA, even transient incubation with EA significantly inhibited LPS-induced secretion of IL-6, IL-10, and nitric oxide (Table 2).

EA Inhibits LPS-Induced HMGB1 Secretion. Preliminary experiments revealed that the concentration of LPS required to reliably induce RAW 264.7 cells to secrete HMGB1 was greater than that required to induced secretion of IL-6, IL-10, or nitric oxide. Accordingly, for these experiments, RAW 264.7 cells were cultured in the absence or presence of 1000 ng/ml LPS. Although Wang et al. (1999) reported detectable levels of HMGB1 in supernatants from RAW 264.7 cells exposed to LPS for only 8 h, in our hands, a longer period of incubation (≥48 h) was required to reliably assay HMGB1 in culture supernatants by Western blot analysis. Therefore, for this series of experiments, we harvested conditioned media after 72 h of incubation in the presence or absence of LPS. To determine whether EA inhibits LPS-induced HMGB1 secretion, graded concentrations of the compound were added to the cultures during the entire period of incubation with LPS. As a positive control for HMGB1 on Western blots, we assayed samples of a whole cell lysate prepared by subjecting RAW 264.7 cells to several freeze-thaw cycles. Whereas unstimulated RAW 264.7 cells did not secrete HMGB1 into the medium, cells incubated with LPS for 72 h released HMGB1 into culture supernatants (Fig. 1A). Coincubation with 15 μM EA partially inhibited LPS-induced HMGB1 secretion and coincubation with 60 μM EA completely inhibited LPS-induced HMGB1 secretion. This effect of EA was not caused by death of the cells, because dead RAW 264.7 cells release large amounts of HMGB1.

In an additional experiment, we pretreated RAW 264.7 cells with control medium or 30 μM EA for various intervals (120, 60, or 0 min) and then incubated the cells with LPS in the absence or presence of EA for 24 h (for determination of nitric oxide release) or 48 h (for determination of HMGB1 secretion). Irrespective of the pretreatment interval used, 30 μM EA inhibited LPS-induced nitric oxide formation and HMGB1 secretion (data not shown).

Transient Exposure to EA Inhibits LPS-Induced HMGB1 Secretion. We sought to determine whether transient exposure of cells to EA leads to persistent inhibition of LPS-induced secretion of HMGB1. Accordingly, RAW 264.7 cells were pretreated with graded concentrations of EA for 1 h, and then the cells were washed extensively before being exposed for 72 h to 1000 ng/ml LPS. As was the case with persistent exposure to EA, even transient incubation with this compound inhibited LPS-induced HMGB1 release, although higher concentrations (≥60 μM) were required (Fig. 1B).

EA Inhibits LPS-Induced HMGB1 Secretion by a Mechanism Other Than Inhibition of NF-κB DNA Binding. We recently reported that EA inhibits activation of the proinflammatory transcription factor NF-κB in LPS-stimulated RAW 264.7 cells (Han et al., 2005). We reasoned that if HMGB1 inhibits LPS-induced HMGB1 secretion by virtue of this mechanism, then another (chemically dissimilar) NF-κB activation inhibitor also should down-regulate secretion of HMGB1 by RAW 264.7 cells incubated with LPS. In this regard, PDTC has been used widely as a pharmacological inhibitor of NF-κB-dependent signaling. PDTC has been shown to inhibit NF-κB DNA binding (Xie et al., 1994) and NF-κB-dependent gene transcription (Woo et al., 2004) in LPS-stimulated RAW 264.7 cells. Accordingly, RAW 264.7 cells were preincubated with vehicle, 30 μM EA, or 100 μM PDTC for 1 h before incubation in the absence or presence of 100 ng/ml LPS for 30 min. As expected, both EA and PDTC inhibited LPS-induced NF-κB DNA binding, as assessed by EMSA, although (at the concentrations tested) PDTC was more effective than EA (Fig. 2A). However, 30 μM EA clearly inhibited LPS-induced HMGB1 secretion, whereas 100 μM PDTC had no effect on HMGB1 release from RAW 264.7 cells incubated for the same interval (48 h) with LPS (Fig. 2B).

Although it is widely used as a pharmacological inhibitor of NF-κB-dependent signaling, PDTC reportedly has other pharmacological effects. Accordingly, we investigated the effects of two other inhibitors of NF-κB DNA binding, namely, the selective IκB kinase-2 inhibitor SC-514 (Kishore et al., 2005) and SN50, a cell-permeable peptide that blocks nuclear translocation of NF-κB (Lin et al., 1995). We verified that both of these agents inhibited LPS-induced NF-κB DNA binding in RAW 264.7 cells (Fig. 3A). However, like PDTC, neither SN-514 (Fig. 3B) nor SN50 (Fig. 3C) inhibited LPS-induced HMGB1 release from RAW 264.7 cells. Collectively, these data support the view that inhibition of HMGB1 release by EA depends on a mechanism other than down-regulation of NF-κB DNA binding.

EA Inhibits LPS-Induced Phosphorylation of ERK1/2. Pharmacological inhibition of signaling mediated by the mitogen-activated protein kinases ERK1/2 has been shown to inhibit HMGB1 secretion by stimulated THP-1 macrophage-like cells (Kalinina et al., 2005). Accordingly, we sought to determine whether EA inhibits LPS-induced phosphorylation in RAW 264.7 cells. The cells were preincubated for 90 min in the absence or presence of graded concentrations of EA. Some cells were stimulated with 10 ng/ml LPS for 15 min, whereas others were incubated for the same period in the absence of LPS. Whole cell extracts were prepared and assayed for total ERK1/2 and phosphorylated ERK1/2 by Western blotting. As expected, stimulation of RAW 264.7 cells with LPS did not affect total ERK1/2 expression, but it resulted in rapid phosphorylation of ERK1/2 (Fig. 4A). Pretreatment with EA at a concentration ≥10 μM inhibited ERK1/2 phosphorylation. EA-mediated inhibition of ERK1/2 phosphorylation was reproducible—it was observed in three replicate experiments. Accordingly, we sought to determine whether HMGB1 release is affected when RAW 264.7 cells are incubated with LPS in the presence of U0126, a potent and specific inhibitor of the upstream kinase that is responsible for ERK1/2 activation (Favata et al., 1998). Although we verified that U0126 inhibited LPS-induced ERK1/2 phosphorylation, this compound had no effect on LPS-induced HMGB1 secretion (Fig. 4B). Therefore, neither inhibition of ERK1/2 activation nor inhibition of NF-κB-dependent signaling seems to be a plausible explanation for the ability of EA to down-regulate HMGB1 release from immunostimulated RAW 264.7 cells.

The Phase II Response Inducers dl-Sulforaphane and Oltipraz Inhibit LPS-Induced HMGB1 Secretion. EA is a substituted phenoxyacetic acid, containing a 2-methylene-1-oxobutyl moiety. Because the exo-methylene group is conjugated with an electron-withdrawing substituent (the 1-oxo group), EA is a fairly potent electrophile that is capable of acting as a Michael acceptor for nucleophiles, including the sulfhydryl groups of cysteine-containing proteins. Many potent electrophiles are capable of inducing increased expression of a characteristic set of proteins, including glutathione S-transferases, UDP-glucuronosyltransferases, and NAD(P)H:quinone oxidoreductase (Talalay et al., 1988). Collectively, these proteins are called phase 2 (detoxifying) enzymes (Talalay, 2000), and their induction depends on binding of trans-activating factor(s) to an antioxidant response element upstream from to the gene transcription start site. EA has been shown to increase reporter gene activity 2-fold in cells stably transfected with an NAD(P)H quinone oxidoreductase antioxidant response element-reporter construct (Ciaccio et al., 1994). EA increases expression of the phase 2 proteins, glutathione S-transferase (Tjalkens et al., 1998) and dihydrodiol dehydrogenase (Shen et al., 1995). In view of these data, we hypothesized that other well studied phase 2 enzyme inducers, such as oltipraz (Buetler et al., 1995) and dl-sulforaphane (Zhang et al., 1994; Gerhuaser et al., 1997), might inhibit LPS-induced HMGB1 release. Accordingly, RAW 264.7 cells were incubated for 48 h in the absence or presence of LPS in the absence or presence of graded concentrations of either oltipraz or dl-sulforaphane. As shown in Fig. 5, both compounds inhibited LPS-induced HMGB1 secretion in a concentration-dependent manner. dl-Sulforaphane was somewhat more potent than oltipraz. At the concentrations tested, neither compound was associated with cytotoxicity as assessed by trypan blue dye exclusion (data not shown). Furthermore, cytotoxicity is associated with increased (not decreased) levels of HMGB1 in supernatants (because the nuclear protein is released from necrotic cells) (Scaffidi et al., 2002).

EA and dl-Sulforaphane Inhibit LPS-Induced Nuclear-to-Cytoplasmic Translocation of HMGB1. Because the primary amino acid sequence of HMGB1 is devoid of a signal peptide, secretion of this protein presumably occurs via a nonclassic secretory pathway. Indeed, when monocytes are activated by exposure to LPS, HMGB1 relocalizes from the nucleus into cytoplasmic organelles that belong to the endolysosomal compartment (Gardella et al., 2002). In activated monocytes, the transfer of HMGB1 from the nucleus to the cytoplasm is mediated by hyperacetylation of critical lysine clusters that are components of nuclear localization signals (Bonaldi et al., 2003). In view of these findings, we sought to determine whether EA or dl-sulforaphane interferes with the relocalization of HMGB1 within immunostimulated murine macrophage-like cells. Accordingly, RAW 264.7 cells were grown on coverslips and cultured for 24 h in the absence or presence of LPS in the absence or presence of either 30 μM EA or 30 μM dl-sulforaphane. The fixed cells were permeabilized and stained with a rabbit anti-HMGB1 antibody and an appropriate secondary antibody before being visualized using a confocal scanning fluorescence microscope. Care was taken to image all of the preparations under exactly the same conditions, so that differences in fluorescence intensity could be unambiguously attributed to differences in immunoreactive HMGB1 content. In unstimulated cells, HMGB1 staining was almost completely localized to the nucleus, except in cells that were undergoing mitosis (Figs. 6, A and B, and 7). Following stimulation with LPS, the cells underwent obvious morphological changes, including spreading of the cytoplasm and formation of pseudopodia (Fig. 6, C and D). In addition, cytoplasmic staining for HMGB1 increased significantly (Figs. 6, C and D, and 7). The effects of LPS on cellular morphology and cytoplasmic localization were dependent on the dose of endotoxin used, being much more apparent when cells were exposed to 1000 ng/ml compared with 10 or 100 ng/ml (data not shown). When cells were stimulated with LPS in the presence of either 30 μM EA (Figs. 6, E and F, and 7) or 30 μM dl-sulforaphane (Figs. 6, G and H, and 7), the redistribution of HMGB1 to the cytoplasm was significantly decreased, and the morphological changes associated with activation were attenuated as well.

Discussion

Oronsky et al. (1969) reported that EA is capable of inhibiting the local inflammatory response elicited in rats by implanting a cotton sponge in subcutaneous tissue. One possible explanation for this effect was provided by Brennan et al. (1993), who reported that 100 μM EA inhibits IL-1-induced activation of the proinflammatory transcription factor NF-κB in EL-4 murine thymoma cells. Recently, our group extended this line of investigation by showing that EA inhibits NF-κB DNA binding as well as NF-κB-dependent gene transcription in LPS-stimulated RAW 264.7 cells (Han et al., 2005). More than one mechanism seems to be responsible for EA-mediated inhibition of NF-κB-dependent signaling, because the compound impaired LPS-induced IκB degradation in RAW 264.7 cells but also impaired DNA binding by p50 homodimers in a cell-free system and inhibited spontaneous DNA binding by p65 in cells transfected with a plasmid encoding wild-type p65 (Han et al., 2005).

To further pursue this line of investigation, we used an in vitro reductionist system (LPS-stimulated RAW 264.7 cells) to examine the effects of EA on the release of several pro- or anti-inflammatory mediators. Transient exposure of the cells to EA down-regulated LPS-induced TNF, IL-6, IL-10, and nitric oxide secretion. Persistent exposure to lower concentrations of EA inhibited IL-6, IL-10, and nitric oxide secretion. Although other NF-κB inhibitors are known to be capable of down-regulating TNF, IL-6, and nitric oxide release from immunostimulated macrophage (or macrophage-like cells) (Makarov et al., 1997; Schow and Joly, 1997; Bondeson et al., 1999; Lim et al., 2004), LPS-induced IL-10 secretion is thought to be NF-κB-independent (Bondeson et al., 1999). Accordingly, the ability of EA to inhibit secretion of the anti-inflammatory cytokine IL-10 suggests that this drug is capable of modulating inflammatory signaling by both NF-κB-dependent and NF-κB-independent mechanisms.

The role of NF-κB-dependent signaling in the secretion of HMGB1 by macrophages is presently unclear. In one recent study, Wang et al. (2004) reported that nicotine and other cholinergic agonists inhibit the release of HMGB1 by LPS-stimulated human macrophages. Nicotine did not affect total intracellular protein levels or LPS-induced phosphorylation of the ERK1/2, c-Jun NH₂-terminal kinase, or p38 mitogen-activated protein kinase pathways. In contrast, nicotine inhibited both LPS-induced NF-κB DNA binding and LPS-induced NF-κB-dependent transcription of a reporter gene, suggesting that this signaling pathway is important for HMGB1 secretion. However, contrary results were obtained in another study that used THP-1 human macrophage-like cells (Kalinina et al., 2005). In this study, HMGB1 secretion was induced by incubating the cells with various proinflammatory cytokines, including TNF, tumor necrosis factor-like weak inducer of apoptosis, CD40L, and interferon-γ. HMGB1 secretion induced by these cytokines was not inhibited by isohelenin, a sesquiterpene lactone inhibitor of NF-κB-dependent signaling. In contrast, wortmannin (an inhibitor of phosphoinositol-3-kinase), PD098059 (an inhibitor of ERK1/2 activation), and bisindolylmaleimide (a protein kinase C inhibitor) all down-regulated cytokine-induced HMGB1 secretion.

In our study, three different pharmacological inhibitors of NF-κB-dependent signaling, PDTC, SN50, and SC-514, all down-regulated LPS-induced NF-κB DNA binding as expected, but they failed to inhibit LPS-induced HMGB1 secretion. Accordingly, our findings suggest that inhibition of HMGB1 release by EA occurs via one or more mechanisms that are NF-κB-independent. Because we showed that EA weakly inhibited LPS-induced phosphorylation of ERK1/2, and a well studied ERK1/2 inhibitor, PD098059, was shown to down-regulate HMGB1 secretion from stimulated macrophages (Kalinina et al., 2005), we considered the possibility that inhibition of ERK1/2 mitogen-activated protein kinase-dependent signaling might be the mechanism responsible for EA-mediated inhibition of LPS-induced HMGB1 secretion. In our hands, however, treating RAW 264.7 cells with U0126, a potent and selective inhibitor of ERK1/2 activation (Favata et al., 1998), failed to block LPS-induced HMGB1 secretion.

Another mechanism is suggested by the recognition that EA is an inducer of so-called phase 2 proteins, such as glutathione S-transferase (Tjalkens et al., 1998) and dihydrodiol dehydrogenase (Shen et al., 1995). EA, like other phase 2 protein inducers, is an electrophilic compound that is capable of serving as a Michael acceptor in reactions involving thiolcontaining proteins or small molecules. The induction of phase 2 proteins by various electrophiles is thought to involve sensing by reactive thiols in a cytosolic protein called Keap1 and downstream signaling by a transcription factor called Nrf2 (Holtzclaw et al., 2004). Although it is not known whether EA activates signaling via the Keap1/Nrf2 pathway, the two other phase 2 inducers we studied, namely, oltipraz and dl-sulforaphane, are known to activate this signal transduction chain (Dinkova-Kostova et al., 2002; Pietsch et al., 2003). Like EA, both of these compounds clearly inhibited LPS-induced HMGB1 secretion by LPS-stimulated RAW 264.7 cells. These data support the notion that activation of the Keap1/Nrf2 signaling pathway is capable of down-regulating active secretion of HMGB1 from immunostimulated macrophages. Furthermore, because oltipraz has been extensively studied in human subjects and shown to be relatively safe (Benson et al., 2005), our findings suggest that this compound, or perhaps other phase 2 protein inducers, might warrant evaluation for the prevention or treatment of inflammatory diseases thought to be mediated by HMGB1.

Our data are insufficient to elucidate the precise mechanism whereby drugs such as EA and dl-sulforaphane inhibit LPS-induced HMGB1 secretion. However, our immunofluorescence images may provide some clues in this regard. A key early step in the secretion of HMGB1 from immunostimulated macrophages is relocalization of the protein from the nucleus into cytoplasmic endolysosomes (Gardella et al., 2002). Consistent with the finding obtained in prior studies using RAW 264.7 cells (Rendon-Mitchell et al., 2003) or human monocytes (Gardella et al., 2002; Wang et al., 2004), we found that incubating RAW 264.7 cells with LPS in the absence of EA or dl-sulforaphane increased the amount of immunoreactive HMGB1 in the cytoplasm. This phenomenon clearly was inhibited by coincubating the cells with EA or dl-sulforaphane. Because the nuclear-to-cytoplasmic relocalization of HMGB1 in immunostimulated cells is thought to be mediated by hyperacetylation of critical lysine clusters in the primary sequence of the protein (Bonaldi et al., 2003), it seems plausible that agents such as EA or dl-sulforaphane interfere with this acetylation process in some way. Additional studies that are beyond the scope of the present manuscript will be needed to pursue this line of investigation.

Although EA was effective as an inhibitor of LPS-induced TNF, IL-6, nitric oxide, and HMGB1 secretion, this drug was not very potent. The typical peak plasma concentration of EA was approximately 30 μM when human subjects were treated 100 mg of the drug by intravenous infusion over 15 min (Lacreta et al., 1994). In our in vitro studies, transient exposure to this concentration of EA was not sufficient to markedly down-regulate secretion of TNF, IL-6, IL-10, nitric oxide, or HMGB1. These data suggest that EA may not be sufficiently potent to be useful in vivo for ameliorating systemic inflammatory conditions. But, by the same token, EA might serve as a reasonable starting point for a series of structure-activity studies that seek to optimize potency with respect to inhibition of cytokine, and particularly, HMGB1 secretion.