Involvement of Toll-like Receptors 2 and 4 in Cellular Activation by High Mobility Group Box 1 Protein*

High mobility group box 1 (HMGB1) protein, originally described as a DNA-binding protein that stabilizes nucleosomes and facilitates transcription, can also be released extracellularly during acute inflammatory responses. Exposure of neutrophils, monocytes, or macrophages to HMGB1 results in increased nuclear translocation of NF-κB and enhanced expression of proinflammatory cytokines. Although the receptor for advanced glycation end products (RAGE) has been shown to interact with HMGB1, other putative HMGB1 receptors are known to exist but have not been characterized. In the present experiments, we explored the role of RAGE, Toll-like receptor (TLR) 2, and TLR 4, as well as associated kinases, in HMGB1-induced cellular activation. Culture of neutrophils or macrophages with HMGB1 produced activation of NF-κB through TLR 4-independent mechanisms. Unlike lipopolysaccharide (LPS), which primarily increased the activity of IKKβ, HMGB1 exposure resulted in activation of both IKKα and IKKβ. Kinases and scaffolding proteins downstream of TLR 2 and TLR 4, but not TLR/interleukin-1 receptor (IL-1R)-independent kinases such as tumor necrosis factor receptor-associated factor 2, were involved in the enhancement of NF-κB-dependent transcription by HMGB1. Transfections with dominant negative constructs demonstrated that TLR 2 and TLR 4 were both involved in HMGB1-induced activation of NF-κB. In contrast, RAGE played only a minor role in macrophage activation by HMGB1. Interactions of HMGB1 with TLR 2 and TLR 4 may provide an explanation for the ability of HMGB1 to generate inflammatory responses that are similar to those initiated by LPS.

HMGB1 1 (formerly HMG1) was originally described as a non-histone, chromatin-associated nuclear protein (1)(2)(3)(4). HMGB1 has a highly conserved sequence among species, with murine HMGB1 differing from the human form by only two amino acids. HMGB1-deficient mice die within a few hours of birth, demonstrating the crucial role of this protein in cellular function. HMGB1 consists of two tandem L-shaped domains, HMGB boxes A and B, each ϳ75 amino acids in length, and a highly acidic carboxyl terminus of 30 amino acids in length.
HMGB1 appears to have two distinct functions in cellular systems. First, it has been shown to be an intracellular regulator of transcription, and, second, HMGB1 can occupy an extracellular role in which it promotes tumor metastasis and inflammation (2)(3)(4)(5)(6)(7)(8)(9). Extracellular HMGB1 has been demonstrated to participate in inflammatory processes, including delayed endotoxin lethality and acute lung injury (10,11). Monocytes and macrophages stimulated by lipopolysaccharide (LPS), tumor necrosis factor (TNF)-␣, or interleukin-1 (IL-1) secrete HMGB1 (5,11). Culture of monocytes with HMGB1 results in the release of TNF-␣, IL-1␣, IL-1␤, IL-1Ra, IL-6, IL-8, macrophage inflammatory protein-1␣, macrophage inflammatory protein-1␤, but not IL-10 or IL-12 (5,11). Production of proinflammatory cytokines after exposure to HMGB1 occurs with delayed kinetics as compared with LPS-induced stimulation. For example, culture of macrophages with LPS results in increases in TNF-␣ that are apparent within less than 1 h, whereas TNF-␣ synthesis following HMGB1 exposure only begins to occur after 2 h and then persists for as long as 8 h (8,11). The signaling mechanisms responsible for the delayed expression of proinflammatory cytokines by HMGB1-stimulated cells remain incompletely explained but appear to involve the p38, ERK, JNK, and Akt kinases and to lead to enhanced nuclear translocation of NF-B (12)(13)(14).
In recent studies (13), we found that the magnitude and kinetics of cytokine expression and nuclear translocation of NF-B after culture of neutrophils with HMGB1 or LPS were similar, suggesting that overlapping mechanisms of cellular activation might be involved. Comparison of gene expression arrays also demonstrated substantial but not complete homology in response to HMGB1 and LPS. Signaling via the TLR 4 receptor is responsible for LPS-induced activation of the IKK kinase complex, including IKK␣ and IKK␤, that then leads to phosphorylation and degradation of IB, nuclear translocation of NF-B, and enhanced expression of proinflammatory cytokine genes whose transcription is dependent on NF-B (15)(16)(17). Although the similarity in gene expression, kinase activation, and patterns in NF-B translocation after exposure to HMGB1 and LPS suggest that similar receptors, including TLR 4, might be involved in cellular activation by the two stimuli, the role of TLR 4 in HMGB1-induced signaling has not been reported. The receptor for advanced glycation end products (RAGE), a multiligand member of the immunoglobulin superfamily of cell sur-face molecules, has been shown to interact with HMGB1 (18 -22). Binding of HMGB1 to RAGE leads to neurite outgrowth and enhanced expression of tissue-type plasminogen activator by transformed macrophages (12,(23)(24)(25)(26)(27)(28). In these cell populations, engagement of RAGE leads to activation of NF-B through a redox-dependent pathway involving Ras (24,29,30). RAGE ligation has been shown to activate ERK1/2, p38, and SAPK/JNK kinases, as well as the small GTPases, Rac and cdc42 (12,24,31,32). In addition to being present on neurites and macrophages, RAGE also exists on monocytes, as well as glioma, endothelial, and smooth muscle cells (23,24,26,29,30,33,34). The recent finding that HMGB1-induced differentiation of erythroleukemia cells is independent of RAGE indicates that additional HMGB1 receptors exist (21). However, such RAGE-independent HMGB1 receptors have not yet been identified. Additionally, no studies have examined the relative importance of RAGE versus other putative HMGB1 receptors.
In the present experiments, we explored the role of RAGE, TLR 2, and TLR 4, as well as associated kinases, in HMGB1induced cellular activation. Our studies demonstrate that RAGE plays only a minor role in macrophage activation by HMGB1, whereas signaling through TLR 2 and TLR 4 appears to be of much greater importance. Interactions of HMGB1 with TLR 2 and TLR 4 may provide an explanation of the ability of HMGB1 to potentiate many inflammatory responses initiated by LPS.

MATERIALS AND METHODS
Reagents-LipofectAMINE 2000, RPMI 1640, Dulbecco's modified Eagle's medium, and penicillin/streptomycin were obtained from Invitrogen Corp. (Carlsbad, CA). Defined fetal bovine serum was purchased from HyClone (Logan, Utah). LPS (from Escherichia coli O111: B4) was obtained from Sigma Chemical Co. (St. Louis, MO). The LPS was re-extracted twice with phenol then precipitated from the aqueous phase to ensure that signaling only occurred through TLR 4 (35). The TLR 2 ligands peptidoglycan (PGN) and tripamitoyl lipopeptide PAM 3 -Cys-Ser-Lys 4 (PAM) were purchased from InvivoGen (San Diego, CA). Poly(dI-dC)⅐poly(dI-dC) was purchased from Amersham Biosciences (Piscataway, NJ). The Coomassie-Plus protein assay reagent and BCA protein assay reagent were obtained from Pierce (Rockford, IL). Antibodies for IKK␣ and IKK␤ were purchased from Upstate Inc. (Lake Placid, NY). IB␣ protein was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). HMGB1 was purified from pig thymus by the method of Sanders (36) and contained less than 10 pg/ml LPS by chromogenic assay (37).
Mice-Male C3H/HeN (Harlan Sprague-Dawley, Indianapolis, IN) and C3H/HeJ (Jackson Laboratories, Bar Harbor, ME) mice were purchased at 6 weeks of age and were maintained in the animal colony at the University of Colorado Health Sciences Center (Denver, CO). Transgenic mice lacking TLR 2 (38, 39) (TLR 2Ϫ/Ϫ) were a gift from Dr. Shizuo Akira. All mice were 7-10 weeks of age when experiments were initiated. Experimental procedures were approved by the University of Colorado Health Sciences Center Institutional Animal Care and Use Committee.
Isolation and Culture of Mouse Neutrophils-Peripheral neutrophils were purified from bone marrow cell suspensions as previously described (40 -42). To obtain the bone marrow cell suspension, the femur and tibia of a mouse were flushed with 5 ml of RPMI 1640/penicillin/ streptomycin, and the cells were passed through a glass wool column. The cell pellets from the bone marrow were resuspended in RPMI 1640/5%-defined fetal calf serum and then incubated with 10 l of primary antibodies specific for the cell surface markers F4/80, CD4, CD45R, CD5, and TER119 for 15 min at 4°C. This custom mixture (Stem Cell Technologies) is specific for T and B cells, red blood cells, monocytes, and macrophages. After 15-min incubation with the antibody mixture, 100 l of antibiotin tetrameric antibody complexes was added for an additional 15 min at 4°C. Following this, 60 l of colloidal magnetic dextran iron particles were added to the suspension and incubated for 15 min at 4°C. The entire cell suspension was then placed into a column surrounded by a magnet. The T cells, B cells, red blood cells, monocytes, and macrophages were captured in the column, allowing the neutrophils to pass through by negative selection methods. Viability, as determined by trypan blue exclusion, was consistently greater than 98%. Neutrophil purity, as determined by Cytospin preparations stained with Wright's stain, was greater than 98%.
Neutrophils were resuspended in RPMI 1640/10%-defined fetal calf serum at a final concentration of 5 ϫ 10 6 cells/ml and stimulated with 1000 ng/ml HMGB1 or 100 ng/ml LPS. Control cultures or those containing HMGB1 were supplemented with polymyxin B (10 g/ml) to block any effects of contaminating endotoxin.
Transient Transfection and Luciferase Reporter Assay-The mouse macrophage cell line, RAW 264.7, was plated on 6-well plates at 5 ϫ 10 5 cells/ml on the day before transfection. Combinations of expression plasmid DNAs (1 g/ml) were transfected by using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions. Duplicate wells were set up for each group. The reporter plasmid used was pNF-Bluc (Clontech, Palo Alto, CA). The plasmids expressing dominant negative mutant proteins were gifts: MyD88-(152-296), ⌬IRAK-1, and IRAK-2- ( The cells were then lysed, and luciferase activity was measured using a luciferase reporter assay kit (Promega) according to the manufacturer's instructions. All of the luciferase assays were repeated at least three times. Representative results are shown for each experiment.
Immune Complex Kinase Assay-Cells were lysed in lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM Na 2 EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 10 g/ml leupeptin) at 5 ϫ 10 6 cells/ml. The lysates (5 ϫ 10 6 cells/ sample) were incubated with 1 g of polyclonal antibodies (either anti-IKK␣ or anti-IKK␤) for 3 h at 4°C, and then protein G-Sepharose beads (Invitrogen) were added for an additional 2 h. The beads were washed three times in lysis buffer and then once in kinase buffer (20 mM Tris-HCl, pH 7.4; 20 mM MgCl 2 ; 2 mM EGTA; 0.5 mM sodium vanadate; 10 mM ␤-glycerophosphate; 1 mM dithiothreitol). The kinase reaction was initiated by the addition of 30 l of kinase buffer containing 10 M ATP, 5 Ci of [ 32 P]ATP, and 1 g of IB␣ (Santa Cruz Biotechnology Inc.) as a substrate; this step was allowed to proceed for 30 min at 30°C. The reaction was terminated by the addition of 5ϫ SDS sample buffer. Samples were boiled and resolved by 10% SDS-polyacrylamide gel electrophoresis, and the fixed gel was then exposed to an x-ray film.
Electrophoretic Mobility Shift Assays-To obtain nuclear extracts, the neutrophils were suspended in lysis buffer containing 10 mM Tris⅐HCl (pH 7.5), 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 0.1 mM Na 3 VO 4 , and 0.1% Triton X-100, and the samples were incubated on ice for 20 min. After cytoplasm was removed from the nuclei by 15 passages through a 25-gauge needle, the nuclei were collected by centrifugation at 5,000 ϫ g for 10 min at 4°C. The pellets were suspended in extraction buffer containing 20 mM Tris⅐HCl (pH 7.5), 1.5 mM MgCl 2 , 420 mM NaCl, 0.2 mM EDTA, 0.1% Triton X-100, 25% glycerol, 0.5 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 0.1 mM Na 3 VO 4 . After a 30-min incubation on ice, the suspension was centrifuged at 14,000 ϫ g for 20 min at 4°C, and the supernatants were collected. The protein concentration in the supernatants was determined using Coomassie Plus protein assay reagent (Pierce). Nuclear extracts (5 g) were incubated at room temperature for 15 min in 20 l of reaction buffer containing 10 mM Tris⅐HCl (pH 7.5), 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, and 4% glycerol with 32 P-end-labeled, double-stranded oligonucleotide probe specific for the B site, 5Ј-GCCATGGGGGGATCCCCGAAG-TCC-3Ј (Geneka Biotechnology) and 1 g of poly(dI-dC)⅐poly(dI-dC). The complexes were resolved on 5% polyacrylamide gels in Tris⅐HCl (pH 8.0)-borate-EDTA buffer at 10 V/cm. Dried gels were exposed with Kodak Biomax MS film (Rochester, NY) for 1-24 h at 70°C. Quantification was performed by image analysis using densitometry (ChemiDoc system; Bio-Rad, Hercules, CA).
Statistical Analysis-Data are presented as mean Ϯ S.E. for each experimental group. Student's t test (for comparisons between LPS-and HMGB1-stimulated cells) was used. p Ͻ 0.05 was considered significant. (13), we found that exposure of neutrophils to HMGB1 increased nuclear accumulation of NF-B to approximately the same level as that seen after stimulation with LPS. Although interactions with TLR 4 are primarily responsible for LPS-induced cellular activation, the involvement of TLR 4 in HMGB1 signaling had not previously been determined. To examine the importance of TLR 4 and TLR 2 in HMGB1-associated NF-B activation, we utilized neutrophils from transgenic mice lacking TLR 2 (TLR 2Ϫ/Ϫ), C3H/HeJ, and control C3H/HeN mice. C3H/HeJ mice lack a functional TLR 4 receptor due to a missense point mutation that results in the substitution of histidine for proline within the cytoplasmic portion of TLR 4 (43,44).

HMGB1 Enhances Nuclear Translocation of NF-B through Mechanisms Independent of TLR 4 -In a previous study
After exposure to HMGB1, nuclear translocation of NF-B was increased to approximately the same degree in neutrophils from C3H/HeN and C3H/HeJ mice (Fig. 1A). In both mouse strains, maximal nuclear accumulation of NF-B was present 15 min after stimulation with HMGB1. In contrast, after culture with LPS, the nuclear content of NF-B was not altered in C3H/HeJ neutrophils, but was increased in neutrophils from C3H/HeN mice (Fig. 1B). Such results demonstrate that receptors other than TLR 4 participate in HMGB1-induced cellular activation.
HMGB1 and LPS both increased nuclear translocation of NF-B in neutrophils from TLR 2 Ϫ/Ϫ mice (Fig. 1C). As expected, exposure of TLR 2 Ϫ/Ϫ neutrophils to the TLR 2-specific ligands PGN and PAM did not produce any alterations in the nuclear concentrations of NF-B. Such results indicate that receptors other than TLR 2 are involved in cellular activation after exposure to HMGB1.
To delineate mechanisms by which HMGB1 leads to enhanced nuclear translocation of NF-B, we examined activation of IKK␣ and IKK␤ in HMGB1-and LPS-stimulated neutrophils. IKK␣ and IKK␤ are catalytically active components of the IB kinase (IKK) complex that, when activated after cellular exposure LPS or other stimuli, can phosphorylate members of the IB family, freeing cytoplasmic NF-B to translocate into the nucleus and initiate transcriptional activity (45)(46)(47). Although IKK␣ is dispensable for NF-B activation induced by mediators such as LPS and IL-1, cells from mice deficient in IKK␤ show impaired NF-B activation and IL-6 production in response to such proinflammatory stimuli (48,49).
To assess whether HMGB1 stimulation affects activation of the IKK complex, we cultured neutrophils from C3H/HeJ and C3H/HeN mice with HMGB1 or LPS and then determined activity of IKK␣ and IKK␤ (Fig. 2). Exposure of C3H/HeN neutrophils to LPS primarily resulted in increased activity of IKK␤, with little effect on IKK␣ ( Fig. 2A). In C3H/HeJ neutrophils, as expected, LPS produced no apparent activation of either IKK␣ or IKK␤. Unlike the findings after LPS exposure, stimulation of C3H/HeN or C3H/HeJ neutrophils with HMGB1 resulted in activation of both IKK␣ and IKK␤ (Fig. 2B). Maximal activity of IKK␣ and IKK␤ was present 15 and 30 min after exposure of neutrophils to HMGB1.
Phosphorylation of IB␣ by IKK␣ or IKK␤ is followed by ubiquitination and subsequent proteosomal degradation (50,51). Because HMGB1 and LPS had differing effects on the activation of IKK␣ and IKK␤, we expected that distinct patterns of IB␣ degradation would be found in cells stimulated with HMGB1 compared with those exposed to LPS. As shown in Fig. 2A, there were transient decreases in IB␣ levels, oc-curring 15 and 30 min after initiation of cultures, in neutrophils stimulated with HMGB1. In contrast, LPS induced more prolonged decreases in IB␣ levels (Fig. 2B).
TLR 2 and TLR 4, but Not RAGE, Are Required for HMGB1induced Activation of NF-B-To better characterize the signal transduction pathways involved in HMGB1-induced cellular activation and NF-B-dependent transcriptional activity, we co-transfected murine macrophage RAW 264.7 cells with a NF-B reporter plasmid, empty plasmid, or dominant negative mutant forms of TLR 2, TLR 4, or RAGE. As shown in Fig. 3A, dominant negative TLR 2 significantly inhibited HMGB1-induced NF-B luciferase activity, but not that induced by LPS. In contrast, transfection of dominant negative TLR 4 reduced NF-B-dependent luciferase activity in cells stimulated with either HMGB1 or LPS. Introduction of dominant negative RAGE, a putative receptor for HMGB1, only minimally decreased NF-B activation by either HMGB1 or LPS (Fig. 3A).
To ensure that the observed effects of HMGB1 on NF-B-dependent expression were not due to contaminating LPS, we cultured RAW 264.7 cells that had been transfected with the NF-B reporter plasmid with HMGB1, LPS, or trypsin-digested HMGB1. As shown in Fig. 3B, the increase in luciferase expression produced by culture with intact HMGB1 was absent when trypsin-treated HMGB1 was used. These findings indicate that contaminating LPS is not responsible for the stimulatory effects of the HMGB1 preparation on NF-B-dependent luciferase expression.
The experiments shown in Fig. 3A indicate that cellular activation induced by HMGB1 primarily occurs through TLR 2 and TLR 4, with RAGE playing a lesser role. To examine cooperative effects of these three putative HMGB1 receptors, as well as the possibility that other receptors might be involved in HMGB1 signaling, we performed double and triple transfections of dominant negative receptor constructs into RAW 264.7 cells co-transfected with a NF-B luciferase reporter plasmid. As shown in Fig. 3C, whereas the combination of dominant negative cytoplasmic RAGE with TLR 2 had only minimal additive effects over that found with blockade of TLR 2 alone, co-transfection of dominant negative TLR 4 with TLR2 further decreased luciferase expression. Transfection with dominant negative TLR 2, TLR 4, and RAGE resulted in additional suppression of luciferase expression compared with blockade of TLR 2 and TLR 4. Such findings are consistent with TLR 2 and TLR 4 being the major receptors for HMGB1 but indicate that RAGE also plays a contributory role. Because a low level of NF-B activation remained present in cells in which TLR 2, TLR 4, and RAGE were all blocked, these results also indicate that additional HMGB1 receptors are likely to be present. Fig. 3A demonstrated important roles for TLR 2 and TLR 4 in HMGB1-induced NF-B activation, we suspected that kinases and scaffolding proteins known to be associated with TLR 2 and TLR 4 and involved in LPS signaling would also participate in pathways activated by HMGB1.
Transfection of dominant negative MyD88 or TIRAP inhibited almost all of the HMGB1-and LPS-induced NF-B activation (Fig. 4). There was also significant inhibition of HMGB1and LPS-induced NF-B activity by dominant negative IRAK-1 and IRAK-2, but about half-maximal NF-B activity remained when IRAK-4 was blocked.
HMGB1-induced NF-B Activity Is Mediated by TRAF6, IKK␣, and IKK␤ but Not TRAF2 and NIK-Because activation of NF-B-dependent transcription in cells stimulated with HMGB1 was dependent on MyD88, TIRAP, and IRAK family members, we assumed that downstream kinases, including TRAF 6, TAK 1, TAB 2, IKK␣, IKK␤, and NIK, would also be involved. IRAK-1 interacts with the downstream adaptor TRAF 6, resulting in formation of a complex that also includes TAK 1 and TAB 2 (16,55). In this multiple protein complex, TRAF 6 presumably facilitates the formation of a novel polyubiquitin chain in collaboration with ubiquitin-conjugating enzyme complex (E2) (56). TAK 1 is a MAPKKK that was originally identified in the transforming growth factor ␤ signaling pathway (57) but also appears to be involved in IL-1-induced phosphorylation of the IKK complex, thereby directly participating in NF-B activation (57,58). TRAF 2 is associated with TNF receptors and is involved with TNF-induced JNK/SAPK activation and anti-apoptotic signaling (59). Unlike TRAF 6, there is no evidence of TRAF 2 activation after cellular exposure to LPS.
Transfection of dominant negative IKK␣ or IKK␤ strongly FIG. 1. HMGB1-induced nuclear translocation of NF-B is not solely dependent on TLR 2 or TLR 4. Nuclear extracts were obtained from neutrophils purified from C3H/HeJ, C3H/HeN, or TLR 2 Ϫ/Ϫ mice and were then stimulated with LPS (100 ng/ml), HMGB1 (1 g/ml), PGN (30 g/ml), or PAM (100 ng/ml) for 1 h. Enhanced nuclear accumulation of NF-B was present in both C3H/HeN and C3H/HeJ neutrophils exposed to HMGB1 (A), but only in cells from C3H/HeN mice cultured with LPS (B). Neutrophils incubated with trypsin-digested HMGB1 (0.05% trypsin carried out overnight at 25°C) for 15 min showed no activation of NF-B (shown as *15 in Fig. 1A). Both HMGB1 and LPS increased nuclear translocation of NF-B in TLR 2 Ϫ/Ϫ neutrophils, whereas the TLR 2-specific ligands PGN and PAM had no effect on nuclear accumulation of NF-B (C). The gels shown are representative of three separate experiments.

FIG. 2. Exposure to HMGB1 produces activation of both IKK␣ and IKK␤.
Cell extracts from C3H/HeN and C3H/HeJ neutrophils were obtained at the indicated times after culture with HMGB1 (A) or LPS (B), immunoprecipitated with anti-IKK␣ or anti-IKK␤ antibodies, and the ability to phosphorylate IB␣ was determined by autoradiography. In contrast to LPS exposure, which primarily resulted in IKK␤ activation, stimulation with HMGB1 activated both IKK␣ and IKK␤. Representative gels are shown. Three additional experiments provided similar results.
inhibited both HMGB1-and LPS-induced NF-B activation (Fig. 5A). TRAF 6 also appeared to play a central role in modulating NF-B-dependent transcription in cells stimulated with HMGB1 or LPS. Several kinases upstream of IKK␣ and IKK␤, but downstream of TRAF 6, appeared to be differentially involved in HMGB1 and LPS signaling. Although TAK 1, TAB 2, and NIK all appeared to participate in HMGB1-induced NF-B transcription, their role was less important than that found after cellular stimulation with LPS (Fig. 5B). As expected, TRAF 2 was not involved in LPS-induced activation of NF-B-dependent transcriptional activity. Similarly, elimination of TRAF 2 activity had minimal effects on NF-B activation in cells exposed to HMGB1.
p38 MAPK Participates in HMGB1-induced NF-B Activation-In previous studies using specific kinase inhibitors (13), we found that p38 was involved in HMGB1-induced enhancement of proinflammatory cytokine production in neutrophils and contributed to HMGB1-associated increases in nuclear translocation of NF-B. It has also been shown that p38 participates in LPS-induced nuclear accumulation of NF-B in neutrophils and is involved in modulating NF-B-dependent transcription in macrophages stimulated with LPS (16,60,61).
Transfection of dominant negative p38␣ MAPK decreased HMGB1-and LPS-induced NF-B activity by ϳ80% (Fig. 6). Inhibition of the upstream kinases MKK3 and MKK6, previously shown to participate in p38 activation (61), also reduced NF-B-dependent transcriptional activity in cells stimulated with HMGB1 or LPS but to a lesser extent than did blockade of p38.

DISCUSSION
The present experiments demonstrate important roles for TLR 2 and TLR 4 in HMGB1-induced activation of myeloid cell populations, with only minimal participation of RAGE. Although RAGE is a major receptor for HMGB1 in neural tissue and some malignant cells (19,(23)(24)(25)31), this receptor appears to be less important in HMGB1 signaling among other cell populations, consistent with the results from our studies. For example, incubation of microvascular endothelial cells with anti-RAGE antibodies only decreased HMGB1-induced IL-8 production by 14% and TNF-␣ production by 17% (12). Similarly, in cultured synovial fluid macrophages, sRAGE decreased HMGB1-induced TNF-␣ release by ϳ40%, implying that the majority of cellular activation by HMGB1 was independent of RAGE (27).
Signaling via the TLR 4 receptor is responsible for LPSinduced activation of the IKK kinase complex (i.e. IKK␣/IKK␤), phosphorylation and degradation of IB, nuclear translocation of NF-B, and enhanced expression of NF-B-dependent proinflammatory cytokine genes (15, 17, 62, 63). Although our re-

FIG. 3. TLR 2 and TLR 4 occupy important roles in HMGB1induced NF-B activation.
In A and C, RAW264.7 cells were cotransfected with pNF-B luciferase reporter plasmid and dominant negative plasmids for TLR 2, TLR 4, RAGE, or the appropriate control plasmid. Cell lysates were obtained after 1 h culture with LPS or HMGB1, then assayed for luciferase activity (LA). The percentage of control LA was calculated using the following equation: percentage of control vector ϭ (mean LA stimulated with dominant negative Ϫ mean LA unstimulated with dominant negative) ϫ 100/(mean LA stimulated with empty vector Ϫ mean LA unstimulated with empty vector). In B, the increase in luciferase activity produced by exposure to intact HMGB1 was eliminated after trypsin digestion, indicating that contaminating LPS was not responsible for the stimulatory effects of HMGB1. Means Ϯ S.E. are shown using results from three independent experiments. *, p Ͻ 0.05 and **, p Ͻ 0.01 versus LPS. sults demonstrate that TLR 4 participates in the activation of neutrophils and macrophages by HMGB1 and that HMGB1induced enhancement of NF-B-dependent transcription involves the IKK complex in a manner similar to LPS, it is unlikely that all of the effects of HMGB1 on cellular activation are initiated solely through interactions with TLR 4. In particular, in the present studies we found that HMGB1-induced nuclear accumulation of NF-B was increased to the same degree in neutrophils from C3H/HeJ mice, which lack functional TLR 4, as in neutrophils from control C3H/HeN mice, demonstrating potent TLR 4-independent activation. In addition, transfection of dominant negative TLR 4 decreased, but did not eliminate, NF-B-dependent transcription in HMGB1stimulated RAW 264.7 macrophages, showing that TLR 4-independent mechanisms are involved in HMGB1-induced cellular activation.
In these experiments, many of the kinases and scaffolding proteins known to play a role in signaling through TLR 2 and TLR 4 were also involved in HMGB1 signaling, consistent with a dominant role for TLR 2 and TLR 4 in HMGB1-induced cellular activation. In particular, inhibition of the function of proximal TLR-associated adaptor proteins, including TIRAP, MyD88, and IRAK-2, as well as kinases, such as IRAK-1 and TRAF 6, decreased NF-B-dependent transcription, whether induced by LPS or by HMGB1, by 80% or more. In contrast, TRAF 2, which is not associated with TLR/IL-1R, but rather with TNFR1 and TNFR2, did not appear to be involved in HMGB1 signaling. Such results, showing that proteins associated with TLR 2 and TLR 4 also participate in an important manner in HMGB1-induced cellular activation, are consistent with experiments in which transfection with dominant negative TLR 4 and TLR 2 constructs also diminished HMGB1associated NF-B activation and support the central role of TLR 2 and TLR 4 in HMGB1 signaling.
Although these studies primarily found similarities in the signaling pathways activated by LPS and HMGB1, there were also a number of differences that may account for the distinct gene expression profiles found in neutrophils activated with LPS as compared with HMGB1. In particular, transfection of dominant negative constructs for the kinase TAK-1 and for the adapter proteins TAB-2 and NIK had greater suppressive effects on LPS-induced NF-B-dependent transcription than that induced by HMGB1. TAK-1 is a MAPKKK and has been shown to participate in several signaling pathways independent of the TLR⅐IL-1R complex (16,55). In particular, TAK-1 was originally identified in the transforming growth factor-␤ activation pathway but has also been demonstrated to be involved in IL-1and LPS-induced phosphorylation of the IKK complex (16,47,55,57,58). Additionally, TAK-1 can phosphorylate MAPK kinases (MKKs) 3, 4, and 6, leading to the activation of p38 and JNK kinases. Of note, in our previous work (13), we found that HMGB1 activated p38 with different kinetics than that found after stimulation with LPS, consistent with distinct influences of HMGB1 and LPS on TAK-1 activation.
The present experiments demonstrate that the p38 MAPK is important in HMGB1-induced enhancement of NF-B activation. The central role of p38 in HMGB1-induced signaling is not surprising, because ligand interaction with all three putative HMGB1 receptors results in p38 activation. Engagement of TLR 2 or TLR 4 results in phosphorylation and activation of p38, which is then important in modulating NF-B transcriptional activity (16,60,64). Similarly, interaction of RAGE with the HMGB1-like protein amphoterin activates p38 (20). Transfection of dominant negative p38␣ reduced LPS-induced NF-B-dependent transcription by almost 80%. A similarly important role for p38␣ in HMGB1-induced cellular activation was found in the present experiments. In our previous studies (13), the proinflammatory effects of HMGB1 in neutrophils primarily appeared to involve the p38 pathway. In those experiments, inhibition of p38 activation significantly decreased nuclear translocation of NF-B and expression of NF-B-dependent cytokines in HMGB1-stimulated neutrophils. Similar findings were reported in endothelial cells where p38 blockade reduced HMGB1-induced IL-8 release by nearly two-thirds (12).
Several studies have shown that incubation of HMGB1 with soluble RAGE can decrease, but not eliminate, the stimulatory effects of HMGB1 on cellular activation (23,27,65). Such findings are not only consistent with those found in our experiments, where RAGE was found to play only a modest role in HMGB1 signaling, but also suggest that the HMGB1 epitopes recognized by RAGE may be distinct from those that interact with TLR 2 or TLR 4. HMGB1 has been demonstrated to interact through its A and B box domains, and possibly through the highly acidic C-terminal domain, with a wide range of proteins, including transcriptional factors, steroid receptors, and viral proteins (66). Using a phage display approach, HMGB1 was shown to recognize more than twelve different peptide sequences. Although several of the peptide sequences associated with both the HMG A and B boxes, others only were isolated with each HMG box. The heterogeneity of peptides that interact with HMGB1 indicates that HMGB1 does not have a highly preferred interaction sequence and is capable of associating with multiple proteins. Such findings may be relevant to the present studies that demonstrate involvement of at least three distinct receptors, i.e. TLR 2, TLR 4, and RAGE, in HMGB1-induced cellular activation.
Interactions of HMGB1 with TLR 2 and TLR 4 may provide an explanation of the ability of HMGB1 to enhance inflammatory responses initiated by LPS. In models of endotoxin-induced shock, HMGB1 has been demonstrated to be a late mediator of lethality and to contribute to the increased levels of circulating and tissue cytokines that are present hours to days after the initial exposure to LPS (5,8,11). HMGB1 is produced and released into the extracellular space with delayed kinetics after exposure of macrophages and monocytes to LPS and is then able to induce further generation of proinflammatory cytokines and chemokines, such as TNF-␣, IL-1␤, and IL-8, in amounts similar to those found after culture of the same cell populations with LPS (5). Engagement of HMGB1 with TLR 2 and TLR 4, with recapitulation of signaling cascades originally activated by LPS, could result in perpetuation of proinflammatory responses that resemble those initiated by LPS. Our results also suggest that interventions able to inhibit intracellular signaling associated with TLR/IL-1R would be effective in reducing the inflammatory actions of HMGB1.