SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

High-mobility group box chromosomal protein 1 (HMGB1) is a protein with both intranuclear functions and extracellular cytokine-like effects. In this report, we study possible candidate receptors for HMGB1 on macrophages (Mφ) and define pathways activated by HMGB1 binding. Bone marrow Mφ were prepared from Dark Agouti (DA) rats and stimulated in vitro with HMGB1. The kinetics of tumour necrosis factor (TNF) production, NO production, activation of p38 mitogen-activated protein kinase (MAPK), p44/42 MAPK- and SAPK/JNK-signalling pathways, nuclear translocation of nuclear factor kappa B (NF-κB) and HMGB1-induced upregulation of major histocompatibility complex (MHC) class II and CD86 were analysed. Mφ from interleukin (IL)-1 receptor type I–/–, Toll-like receptor 2 (TLR2–/–) and RAGE–/– mice were used to investigate the role of these receptors in HMGB1 signalling. HMGB1 induced TNF and NO production by Mφ, phosphorylation of all investigated MAP kinase pathways and NF-κB translocation, and expression of MHC class II was increased. Mφ from RAGE–/– mice produced significantly lower amounts of TNF, IL-1β and IL-6, while IL-1RI–/– and TLR2–/– Mφ produced cytokine levels comparable with wildtype controls in response to HMGB1 stimulation. We conclude that HMGB1 has the potential to induce a proinflammatory phenotype in Mφ, with RAGE as the major activation-inducing receptor.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

High-mobility group box chromosomal protein 1 (HMGB1, previously called HMG-1 or amphoterin) [1] is an abundant nuclear protein expressed in all eucaryotic cells [2–4]. As an intranuclear protein, HMGB1 facilitates gene transcription by enabling the nicking of DNA and by stabilizing nucleosomes [3, 5–7]. In addition to its intranuclear activities, HMGB1 has previously been demonstrated to exert extracellular proinflammatory activities. It is actively secreted by monocytes and macrophages (Mφ) as well as by pituicytes in response to proinflammatory stimuli such as lipopolysaccharide (LPS), tumour necrosis factor (TNF) and interleukin (IL)-1 via a nonclassical, vesicle-mediated pathway [8, 9]. Once extracellularly released, HMGB1 stimulates the production of TNF and IL-1 by human peripheral blood mononuclear cells (PBMC) [10]. In in vitro assays, HMGB1-induced TNF production is both biphasic and delayed in comparison to TNF production following LPS activation [10].

It was recently reported that HMGB1 is passively released from the nuclei of necrotic, but not from apoptotic cells [11, 12]. In healthy cells, HMGB1 is loosely bound to chromatin, but leaks out of cells when membrane integrity is lost in necrotic or chemically permeabilized cells. In contrast, HMGB1 is tightly bound to chromatin in apoptotic cells due to a generalized underacetylation of histones. In apoptotic cells undergoing secondary necrosis, HMGB1 remains tightly bound to chromatin and does not leak out of cells. HMGB1-deficient necrotic fibroblasts had a greatly reduced ability to activate monocytes to TNF production and to promote inflammation in vitro. Antibodies to HMGB1 reduced inflammation in a liver necrosis model induced by acetaminophen [12]. HMGB1 might thus be an important link between unprogrammed cell death and inflammation, acting as an endogenous danger signal.

Extracellular and cytoplasmic HMGB1 has been detected in both acute inflammatory conditions and during chronic inflammation [13]. Through its stimulation of multiple proinflammatory cytokine syntheses, HMGB1 mediates endotoxin lethality [14, 15], acute lung injury [16] and haemorrhagic shock [17]. We have previously demonstrated that non-nuclear HMGB1 is overexpressed in both human and experimental arthritis [18].

HMGB1 can also mediate smooth muscle chemotaxis [19], neurite outgrowth [20] and proinflammatory responses of endothelial cells [21, 22]. For neurites and certain tumour cells, HMGB1 exerts these effects by binding to RAGE (the receptor of advanced glycation end products) [23, 24]. In such cells, the interaction of HMGB1 with RAGE activates several intracellular signal-transduction pathways including mitogen-activated protein kinases, Cdc42, Rac and nuclear translocation of nuclear factor kappa B (NF-κB) [20, 25], the transcription factor classically linked to inflammatory processes. In endothelial cells, HMGB1 induces a transient phosphorylation of MAP kinases, expression of adhesion molecules (e.g. ICAM-1 and VCAM-1) and regulates fibrinolysis, thereby further amplifying the inflammatory response [22]. In one report, it has been suggested that HMGB1 can bind to the IL-1R type I and could thereby activate T cells in a thymocyte-proliferation assay, IL-1RI binding of HMGB1 being estimated using biosensor measurements [26]. In another report, it was suggested that HMGB1 can also bind to Toll-like receptor 2 (TLR2) [27]. Whether HMGB1 binding to these disparate receptors is comparable is presently unknown.

Extracellular HMGB1 is known to induce production of proinflammatory cytokines by human peripheral blood monocytes [10] and by murine Mφ-like (RAW 264.7) cell cultures [28], although the mechanisms leading to this cytokine production are still unclear. We thus set out to investigate possible candidate receptors (including RAGE) for HMGB1 on Mφ and to define pathways activated by HMGB1 after binding to cell-surface receptors. To our knowledge, this is the first report characterizing the effects of HMGB1 on rat and mouse primary Mφ, and we demonstrate that RAGE is the major functional candidate receptor mediating HMGB1-induced activation of rodent Mφ.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Animals.  Female Dark Agouti (DA) rats, male C57BL/6 mice, male B6.129S7-Il 1r1 (IL-1RI –/–) mice, male RAGE–/– mice (129/SW/B6) and male TLR2–/– mice were used in our experiments. DA rats were originally obtained from the Zentralinstitut fur Versuchstierzucht (Hannover, Germany) and were kept and bred at the animal unit at Karolinska Hospital; C57BL/6 and B6.129S7-Il 1r1 (IL-1RI–/–) mice were from Jackson Laboratories (Bar Harbor, ME, USA) and RAGE–/– mice originally obtained from Dr Bernd Arnold, German Cancer Research Center, Heidelberg, Germany; TLR2–/– mice were kindly provided by Dr S. Akira (Osaka University, Osaka, Japan) to the Microbiology and Tumor Biology Center (MTC), Karolinska Institute animal house facility. All mice were kept and bred at MTC, where the conditions are specific pathogen-free, and they had free access to chow and water. Ethical approval was granted by the Ethical Committee of Stockholm North, Sweden.

Macrophage cell culture and harvesting.  Femoral BM cells were collected and single cell suspensions were prepared and resuspended in DMEM medium (Life Technologies, Paisley, Scotland, UK) supplemented with 20% heat-inactivated fetal calf serum, 1 mm sodium pyruvate, 100 U/ml penicillin, 100 µg/ml streptomycin, 2 mm l-glutamine and 2-ME (all reagents from Life Technologies) and 20% L929 cell-line supernatant. 2 × 106 cells/ml were cultured at 37 °C in a humidified incubator with 5% CO2 for 6 days, and an additional 3 days without L929 supernatant. Cells were harvested by adding prewarmed (37 °C) trypsin (Life Technologies) for 10 min followed by washing at 140×g, 6 min at 4 °C.

Reagents.  Recombinant rat HMGB1 (100% homology to mouse HMGB1), referred to herein as HMGB1, was produced and purified as previously described [15]. The protein eluates were passed over a polymyxin B column to remove any contaminating LPS. The endotoxin content of proteins used in cell cultures was measured by using chromogenic Limulus amoebocyte lysate assay (Bio Whittaker, Walkersville, MD, USA). Endotoxin content in HMGB1 preparation was 192 pg/µg protein. To further exclude endotoxin contamination in stimulation experiments, polymyxin B sulfate (Sigma–Aldrich Chemie, Steinhem, Germany) was added to cultures stimulated with HMGB1 at a final concentration of 20 µg/ml. LPS used in all our stimulation experiments was from E. coli strain 055:B5*G (Sigma–Aldrich Chemie, Schnelldorf, Germany). A concentration of 10 µg/ml was used in all experiments.

Enzyme-linked immunosorbent assay (ELISA).  Supernatant from rat Mφ which had been incubated at a concentration 105 cells/ml with or without stimuli (HMGB1 10 µg/ml or LPS 10 µg/ml) for 1, 2, 4, 6, 8, 10, 12, 24, 36 or 48 h, were run in triplicate and analysed using a commercial ELISA kit (Rat TNF, Eli-pair, Biosource, Camarillo, CA, USA). The procedure was performed as recommended by the manufacturer, and absorbance was measured at 450 nm using a microplate reader (Multiscan MS™, Labsystems, Helsinki, Finland).

Supernatants from mouse Mφ cultured as above and stimulated with HMGB1 for 8 h were assayed for IL-1β and IL-6 production by ELISA (Biosource) according to the manufacturer's instructions.

Quantification of nitrite in culture supernatant.  Nitric oxide released from Mφ was determined by the accumulation of nitrites in supernatant. Measurement of NO2 using the Griess reagent provides a surrogate marker and quantitative indicator of nitric oxide production. Supernatants from Mφ which had been incubated at a concentration 105 cells/ml with or without stimuli (HMGB1 10 µg/ml or LPS 10 µg/ml) were analysed at different time points (4, 8, 24, 48, 72, 96, 120 or 144 h). Samples were plated in triplicate and mixed with an equal volume of freshly prepared Griess reagent (modified) (Sigma–Aldrich, St. Louis, MO, USA) for 15 min at room temperature (RT). Absorbance was measured at 540 nm using a microplate reader (Multiscan MS™), and results were quantified by reference to a standard curve prepared from sodium nitrite (Sigma–Aldrich Co, USA) ranging from 0 to 125 µm.

In vitro analysis of phosphorylation of p38 MAPK, p44/42 MAPK and SAPK/JNK.  Mφ were stimulated with HMGB1 (10 µg/ml) or with LPS (10 µg/ml) for 5, 10, 15 or 60 min. Supernatant was aspirated from Mφ cultures, and the cells were washed with sterile phosphate-buffered saline (PBS) (Life Technologies) and lysed by adding 300 µl of boiling Laemmli sodium dodecyl sulfate (SDS) sample buffer (Bio-Rad Laboratories, Hercules, CA, USA) containing sodium vanadate (final concentration 2.5 mm/ml) and protease inhibitor cocktail (Complete™, Roche Diagnostics, Mannheim, Germany). Cell lysates were immediately aspirated from plates and kept on ice. Samples were sonicated for 30 s, boiled for 5 min, cooled on ice and centrifuged (5000 g, 4 °C, 5 min). Fifty microlitres of each sample was loaded in 7.5 (p44/22, SAPK/JNK) or 10% (p38) SDS-polyacrylamide gel electrophoresis (PAGE) gels. After electrotransfer to nitrocellulose membranes (BioTrace®NT, PN 66485, Pall Life Sciences, Ann Arbor, MI, USA) for 1 h at 100 V, the membranes were first washed with PBS for 5 min at RT, incubated for 1 h in blocking buffer (PBS-0.05% Tween-20 with 5% w/v nonfat dry milk) and washed three times for 10 min each with PBS-0.05% Tween-20. Membranes were incubated with primary antibody (antiphospho-p38 MAP kinase 1/1000, cat. no. 9211; antiphospho-p44/42 1/1000, cat. no. 9101 or antiphospho-SAPK/JNK 1/500, cat. no. 9251, all antibodies from Cell Signaling Technology, Beverly, MA, USA) overnight at 4 °C, washed and incubated with secondary antibody (HRP-linked antirabbit Ig, NA 934, Amersham Pharmacia Biotech, Uppsala, Sweden), diluted 1/2000, for 1 h (RT). Detection of protein was performed using ECL-chemiluminescence (Amersham Pharmacia Biotech) and the membranes were exposed to X-ray film (Hyperfilm ECL, Amersham Pharmacia Biotech). X-ray films were analysed using a computerized image analysis program (Kodak 1D 3.5).

Analysis of NF-κB translocation.  Mφ (5 × 105/ml) were seeded on glass coverslips in wells of 24-well plates and cultured for 48 h. After stimulation with HMGB1 (10 µg/ml), coverslips were washed three times in PBS, fixed with 4% paraformaldehyde solution for 10 min in the dark and washed twice. Cells were permeabilized using 0.1% saponin in EBSS, blocked and anti-NF-κB antibody (p65, rabbit polyclonal IgG, Santa Cruz Biotechnology, Santa Cruz, CA, USA, dilution 1/1000) applied overnight at 4 °C. Following further washing, secondary antibody (Oregon green-conjugated goat antirabbit IgG, Molecular Probes, Eugene, OR, USA, dilution 1/500) was applied for 30 min at 37 °C before nuclear staining with DAPI, diluted 1/1000 in PBS. After final washing, coverslips were mounted in Mowiol (Calbiochem, San Diego, CA, USA) before microscopical analysis.

Immunohistochemical analysis of MHC class II and CD86 upregulation.  Mφ were cultured on 8-well glass slides, 5 × 104 cells/well and stimulated for 48 h with HMGB1 (10 µg/ml), LPS (10 µg/ml), LPS (10 µg/ml) + interferon (IFN)-γ (100 U/ml) (recombinant rat IFN-γ, a kind gift from Dr Peter van der Meide, Utrecht University, Utrecht, the Netherlands) or medium alone. Mφ were washed with PBS and fixed in ice-cold acetone, 30 s in 50% acetone followed by 3 min in 100% acetone. Slides were allowed to dry and then soaked for 10 min in PBS. Endogenous peroxidase activity was blocked with 1% hydrogen peroxide and 0.1% sodium azide for 30 min in darkness at RT. Slides were then washed three times in PBS, and incubated at RT with 2% normal horse serum for 30 min to reduce background signals. After three additional washes in PBS, the slides were incubated overnight at 4 °C with the primary antibody [mouse-antirat MHC class II (OX6), originally obtained from ATCC as hybridoma and cultured and purified in our laboratory, dilution 0.3 µg/ml or rabbit-antirat CD86, Santa Cruz Biotechnology, dilution 1/200] diluted in PBS with 2% normal rat serum (NRS). Slides were next washed three times and incubated for 1 h at RT with a biotinylated horse antimouse IgG antibody diluted 1/80 or a biotinylated donkey antirabbit (Jackson Immunoresearch Laboratories, West Grove, PA, USA) diluted 1/500 in PBS with 2% NRS. After three washes with PBS, the slides were incubated for 1 h at RT with avidin–biotin complex, diluted 1/100 (Vectastain peroxidase standard kit, Vector Laboratories, Burlingame, CA, USA). After three final washes, the substrate diaminobenzidine was added (Peroxidase substrate kit, Vector Laboratories). The reaction was stopped after 10 min by addition of deionized water, after which the sections were quickly counterstained with Mayer's hematoxylin. Slides were finally air dried and mounted with buffered glycerol. Number of cells staining positively for each group was counted on three separate fields containing approximately 400 cells each.

Western immunoblot analysis of TNF production in vitro.  Western immunoblotting analyses of Mφ lysates was performed as described using as primary antibody a polyclonal Ag-affinity-purified rabbit antirat TNF (cat. no. CT 061, U-CyTech BV, Utrecht University, the Netherlands), final concentration 5 µg/ml, and as secondary antibody HRP-linked antirabbit Ig (NA 934, Amersham Pharmacia Biotech, UK), diluted 1/2000, for 1 h (RT).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

HMGB1 induces TNF production by rat macrophages

To investigate the kinetics of HMGB1-induced TNF production, DA rat BM Mφ were stimulated with HMGB1 in the presence of polymyxin B. LPS and medium alone were used as positive and negative controls, respectively. Mφ supernatants were harvested at multiple timepoints, and the levels of TNF were measured by ELISA. Within 2 h of incubation with either HMGB1 or LPS, levels of TNF were significantly increased compared to that in nonstimulated controls. HMGB1-induced TNF production reached a first peak at 6 h, then shortly declined, reached a second peak at 10 h and then gradually started declining again (Fig. 1). These results exhibit a somewhat faster kinetics compared to that in our earlier study with human PBMC, in which HMGB1-induced TNF production peaked at 8 h [10]. LPS-induced TNF release was similar to HMGB1-induced release during the whole experiment, reaching peak levels at 8 h, after which it declined, reaching the second peak after 12 h. Thus, HMGB1 is also a stimulator of cytokine production in rat Mφ as well as in human PBMC.

image

Figure 1. The production of tumour necrosis factor (TNF) by Dark Agouti (DA) BM macrophages (Mφ). Macrophages were stimulated with high-mobility group box chromosomal protein 1 (HMGB1) (10 µg/ml), and the concentration of TNF (ng/ml) in cell-culture supernatant was measured by enzyme-linked immunosorbent assay after various time points (0, 1, 2, 4, 6, 8, 10, 12, 24, 36 and 48 h). Lipopolysaccharide (LPS) was used as a positive control. The results are from one representative experiment (of three performed with similar results). LPS-induced TNF production was weaker as compared to TNF production after HMGB1 stimulation. Bars represent standard deviation.

Download figure to PowerPoint

Mφ produce NO in response to HMGB1 stimulation

Reactive nitrogen intermediates produced by activated phagocytic cells are potential antimicrobial and proinflammatory agents. Activated Mφ are known to produce iNOS, an enzyme leading to NO production, in response to stimulation with bacterial components and cytokines. To investigate whether HMGB1 induces iNOS and NO production by Mφ, the levels of NO2, a stable metabolite of NO, were measured after various time points following stimulation with HMGB1, LPS or medium. Increased levels compared to controls were detected 8 h after stimulation with LPS. Both HMGB1- and LPS-induced production reached peak levels at 96 h poststimulation (Fig. 2). Hence, HMGB1 can amplify the inflammatory response not only through cytokine induction, but also by stimulating the production of reactive nitrogen intermediates.

image

Figure 2. Nitrite release by high-mobility group box chromosomal protein 1 (HMGB1)-stimulated Dark Agouti BM macrophages (Mφ). release by HMGB1-stimulated Mφ was recorded at different time points (4–144 h) using the Griess reagent. Lipopolysaccharide (LPS) and medium were used as positive and negative controls, respectively. The results are from one representative experiment (of three performed with similar results). LPS-induced release was faster and higher as compared to HMGB1. The peak level of release was reached at 96 h with both stimuli. Bars represent standard deviation.

Download figure to PowerPoint

Transient phosphorylation of three different MAP kinases in rat Mφ is induced by HMGB1

One of the most important events triggered by the majority of proinflammatory cytokines and by bacterial components is the activation of MAP kinases (mitogen-activated protein kinases). It has previously been reported that these intracellular pathways are employed by HMGB1/amphoterin when leading to neurite outgrowth and activation of endothelial cells [20, 22]. Activation of these intracellular signalling pathways can also lead to release of TNF and NO, the two agents we demonstrated to be induced by HMGB1 stimulation. We investigated the activation of three different MAP kinases, p38 MAPK, P44/42 MAPK and SAPK/JNK in rat Mφ. Transient phosphorylation of these molecules was determined by Western immunoblotting and image analysis (Fig. 3).

image

Figure 3. High-mobility group box chromosomal protein 1 (HMGB1)induces a transient phosphorylation of three different mitogen-activated protein kinase (MAP) kinases. Macrophages (Mφ) derived from Dark Agouti rat BM were stimulated with HMGB1 (10 µg/ml), lipopolysaccharide (LPS) (10 µg/ml) or medium alone (control) and the phosphorylation of p38 MAPK (A), p44/42 MAPK (B) and SAPK/JNK (C) was recorded at various timepoints (5, 10, 15 or 60 min) by Western immunoblotting technique by using antibodies against the phosphorylated forms of MAP kinases. The grade of phosphorylation is expressed as net intensity [net intensity = (mean intensity − background intensity) × area] as analysed by image analysis. Phospho-p38 MAPK (A) and phospho-SAPK/JNK (C) were weakly expressed at baseline and then upregulated. HMGB1-stimulated Mφ exhibited weaker phosphorylation as compared to the cells stimulated with LPS. Phospho-p44/42 (B) was expressed at baseline and strongly upregulated after 15 min (LPS-stimulated cells) and 60 min (HMGB1-stimulated cells). Data demonstrate one of three separate experiments with consistent results.

Download figure to PowerPoint

Weak phosphorylation of p38 MAPK was detected 5 min after incubation with HMGB1, decreasing slightly after 10 min and reaching a plateau after 15 and 60 min (Fig. 3A). LPS-stimulated cells exhibited a similar kinetics as for HMGB1-stimulated cells, with phosphorylation of p38 MAPK already after 5 min, a decrease after 10 min, then an increase again and overall significant higher peak levels compared to HMGB1 stimulation. A weak phosphorylation of p44/42 MAPK was recorded in nonstimulated control samples. HMGB1-stimulated cells induced phosphorylation similar to control samples at 5, 10 and 15 min, but this was increased after 60 min (Fig. 3B). LPS-stimulated cells exhibited a faster kinetics compared to HMGB1-induced activation, with increased phosphorylation after 15 min. Phosphorylation of SAPK/JNK was detected after 10 min and peaked 15 min after stimulation with HMGB1 (Fig. 3C), then clearly declined after 60 min. In contrast, when stimulated with LPS, SAPK/JNK phosphorylation could be recorded after 5 min and was still evident at 60 min.

HMGB1 induces nuclear NF-κB translocation in rat Mφ

The translocation of NF-κB after stimulation with HMGB1 was investigated by immunohistochemistry. The nuclear translocation of NF-κB is triggered by proinflammatory cytokines, and it has previously been reported that HMGB1 activates NF-κB in endothelial cells [21, 22]. Staining of NF-κB in nonstimulated Mφ was primarily restricted to the cell cytoplasm (Fig. 4A). In contrast, nuclear translocation of NF-κB was apparent after 30 min following stimulation with HMGB1 (Fig. 4B).

image

Figure 4. Nuclear translocation of nuclear factor (NF-κB) is induced by high-mobility group box chromosomal protein 1 (HMGB1). The location of NF-κB was analysed by immunofluorescence in unstimulated (A) and HMGB1-stimulated (10 µg/ml) (B) macrophage (Mφ) after 30 min. Unstimulated Mφ reveal cytoplasmic staining, while intranuclear translocation of NF-κB is evident in stimulated cells.

Download figure to PowerPoint

HMGB1 upregulates the expression of MHC class II molecules on the surface of rat Mφ

Immunohistochemical analysis was performed following stimulation of rat Mφ with HMGB1 to assess levels of expression of MHC class II molecules. After 48 h of stimulation, a two-fold increase in the number of Mφ-expressing MHC class II molecules (15 ± 1.3%) was apparent compared to unstimulated control cells (8 ± 3.2%). However, HMGB1 proved to be a significantly weaker stimulator of MHC class II expression than the combination of LPS and IFN-γ, a well-documented upregulator of MHC molecule surface expression (52 ± 6.7%). Analysis of the expression of CD86 (B7-2) revealed occasional cells staining positively, but obvious upregulation of either the number or the intensity of CD86-expressing cells following HMGB1 stimulation was not apparent (data not included).

The role of RAGE and IL-1R in HMGB1-induced Mφ-activation

The importance of two receptors, RAGE and IL-1RI, for induction of intracellular signalling triggered by HMGB1 was investigated. BM Mφ from RAGE–/– mice, IL-1R type I–/– mice and their respective wildtype (wt) littermates were stimulated in vitro with HMGB1. The phosphorylation of all three MAP kinases after 30 min of stimulation was investigated by Western immunoblotting. No significant differences were recorded when either HMGB1-stimulated RAGE–/– or IL-1RI–/– Mφ were compared with their respective wt controls concerning phosphorylation of p38, p44/42 or SAPK/JNK (Fig. 5).

image

Figure 5. High-mobility group box chromosomal protein 1 (HMGB1) induces mitogen-activated protein (MAP) kinase phosphorylation in RAGE–/– and IL-1RI–/– mouse macrophage (Mφ). Mφ derived from RAGE–/– and IL-1R–/– mouse BM were stimulated with HMGB1 (10 µg/ml) or medium alone (control) and the phosphorylation of p38 MAPK, p44/42 MAPK and SAPK/JNK was recorded after 30 min by Western Immunoblotting using antibodies against the phosphorylated forms of MAP kinases. The grade of phosphorylation is expressed as net intensity analysed by image analysis.

Download figure to PowerPoint

HMGB1 induces cytokine production in RAGE–/–, IL-1R type I–/– and TLR2–/–

In contrast, significant differences were detected when the production of TNF, IL-1β and IL-6 was analysed. TLR2–/– mice and wt controls (C57BL/6) were additionally used in these analyses. Both RAGE–/– Mφ and their wt controls produced TNF, IL-1β and IL-6 after 8 h of stimulation with HMGB1. Compared to wt Mφ, the levels of TNF, IL-1β and IL-6 production by RAGE–/– Mφ were 70% lower, albeit still significantly higher than nonstimulated wt controls (Table 1). Conversely, IL-1R type I–/– Mφ exhibited only marginally higher levels of TNF expression compared to that in wt cells (Table 1), and TNF expression was not reduced in TLR2–/– cells.

Table 1.  Production of TNF, IL-1β and IL-6 after 8 h of stimulation with HMGB1 in different mouse strains
 Mouse strain
ParameterC57BL/6IL-1RI KORAGE KOTLR2 KO
  1. IL, interleukin; KO, knockout; ND, not determined; TNF, tumour necrosis factor; TLR, Toll-like receptor.

  2. Data are represented as percentage. Data for TNF production by RAGE-deficient macrophages after stimulation with high-mobility group box chromosomal protein 1 (10 µg/ml) are expressed as a net intensity [net intensity = (mean intensity − mean background) × area].

  3. IL-1β and IL-6 production are analysed by enzyme-linked immunosorbent assay.

TNF production after 8 h10012231100
IL-1β production after 8 h100ND23ND
IL-6 production after 8 h100ND42ND

Taken together, these results indicate that despite HMGB1-induced Mφ activation as indicated by phosphorylation of MAP kinases in all mouse strains, the receptor interaction and functional consequences, i.e. cytokine production, varied depending on the receptor interacting with HMGB1, lack of RAGE greatly reducing the functional readout.

In all experiments with BM Mφ, LPS stimulation was used as a positive control. LPS led to stimulation of knockout and wt Mφ in each assay, there being no significant differences between any of the knockout mice and their wt for any of the signalling and cytokine assays conducted (data not included). This indicates that differences in HMGB1-mediated activation of Mφ were significant.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

In addition to mediating many intranuclear functions [6], HMGB1 has also been described as a cytokine [14, 29, 30] and can be actively secreted from several cell types [8, 14]. We have previously reported that HMGB1 specifically stimulates synthesis of proinflammatory cytokines in human monocytes [10]. Monocytes and Mφ play a central role in both innate and adaptive immunity when acting as phagocytes and antigen-presenting cells. Herein, we address the question of how HMGB1 stimulates primary rodent Mφin vitro, specifically concentrating on the role of different candidate receptors and intracellular signalling pathways in the cytokine- and NO-inducing effects of HMGB1.

The outcome of the present study demonstrates that stimulation of BM-derived rat Mφ with recombinant HMGB1 in vitro leads to production of TNF and NO, two important inflammatory mediators. BM-derived rat Mφ were selected for these studies as they represent a very homologous cell population with little or no prior activation (compared to peritoneal Mφ which are usually heterogeneous and partially activated on removal). When stimulated with HMGB1, TNF concentrations measured in cell-culture medium, reached peak levels after 6 h and slowly started to decline after 10 h. Positive controls (LPS-stimulated Mφ) exhibited similar kinetics compared to HMGB1-stimulated cells and induced lower TNF levels. When compared to our earlier study reporting that HMGB1 induces peak TNF production by human monocytes after 8 h [10], our results exhibit faster kinetics. The reason for this difference is unclear, but it is likely that the sensitivity of rodent Mφ to HMGB1 differs from that of human monocytes. As polymyxin B, an agent neutralizing endotoxin, was added to all cultures stimulated with HMGB1, it is unlikely that the TNF production is induced by LPS contamination. The endotoxin content of HMGB1 used in our cell-stimulation experiments was also measured to further exclude the stimulatory effect of contaminating LPS.

Increased levels of NO2, a stable metabolite for NO and a widely used parameter for production of iNOS and NO, were detectable after HMGB1 stimulation of Mφ. HMGB1 has previously been reported to be a late mediator of endotoxin lethality in mice. It is released by cultured Mφ-like RAW 264.7 cells more than 8 h after stimulation with endotoxin [14]. Interestingly, iNOS has been reported to play a damaging role in several models of endotoxin-induced hepatic injury [31]. HMGB1 might thus be the link coupling endotoxin-induced high iNOS levels in these experimental models. High levels of HMGB1 have also been reported during haemorrhagic shock, another disease associated with increased iNOS levels [17, 32].

As it has previously been demonstrated that HMGB1 induces a transient phosphorylation of MAP kinases and a nuclear translocation of NF-κB in outgrowing neurites, certain tumour cells and human microvascular endothelial cells [20, 22, 24], our findings that three different MAP kinases (p38 MAPK, p44/42 MAPK and SAPK/JNK) were activated and that NF-κB was translocated from the cytoplasm to the nucleus in Mφ following HMGB1 stimulation demonstrate that similar activation pathways are engaged in Mφ as in the previously reported cell types. LPS-stimulated cells exhibited faster kinetics than HMGB1-stimulated cells concerning both the phosphorylation of MAP kinases and translocation of NF-κB. These results could provide an explanation for our earlier findings describing the delay of HMGB1-induced TNF production compared to that following LPS stimulation in human monocytes [10]. The knowledge that HMGB1 induces MAPK activation in Mφ might also offer new therapeutic possibilities through use of MAPK-blocking agents, inhibiting the HMGB1-induced cytokine cascade and NO production.

MHC class II molecules expressed on the surface of Ag-presenting cells have a key role in presenting exogenous Ag to T cells. Here, we report that HMGB1 induces a two-fold increase in the surface expression of MHC class II compared to unstimulated control Mφ. Thus, HMGB1 induces a proinflammatory phenotype in Mφ not only by stimulating cytokine release, but also by increasing the capacity of Mφ to present antigen.

RAGE is reported as a receptor for HMGB1 on endothelial cells, neurites and certain tumour cells [20, 22, 24]. Treatment with anti-RAGE antibodies suppresses tumour growth, metastasis [24] and in part the pro-inflammatory activity of HMGB1 on endothelial cells [21]. Findings presented in this report indicate that RAGE is the major functional receptor mediating the proinflammatory effects of HMGB1 in rodent Mφ. However, in our experiments, RAGE-deficient Mφ stimulated by HMGB1 still produced significant amounts of TNF, indicating the possibility that Mφ express other receptors that are ligands for HMGB1.

It has previously been reported in one article that HMGB1 is associated with IL-1 activity. In the same study, the affinity of IL-1RI for HMGB1 as measured by biosensor studies was described [26]. When we utilized IL-1RI-deficient Mφ, TNF production following HMGB1 stimulation was higher than that in wt controls. These findings indicate that through an unknown mechanism, IL-1RI-deficient mice are more sensitive to the cytokine-stimulating effect of HMGB1 than are wt controls. Thus, we conclude that IL-1RI is not a major receptor for HMGB1 on Mφ leading to cytokine production. HMGB1 has two separate, characteristic DNA-binding domains referred to as HMG A box and B box, respectively, and the B box has recently been characterized as the major cytokine-inducing domain of HMGB1 [27]. It is possible that the affinity of IL-1RI to the B box domain is low, and that the minor cytokine-inducing A box domain binds to the receptor. Additionally, while TLR2 has been reported as an alternative receptor for HMGB1 [27], we could not detect any reduced stimulation of TLR2–/– Mφ following HMGB1 stimulation. These results indicate that as long as RAGE is present on a cell surface, then efficient HMGB1-induced activation is possible.

All three MAP kinases investigated were also phosphorylated after HMGB1 stimulation in RAGE- and IL-1RI-deficient Mφ. As it appears that there are several receptors mediating the cytokine-inducing effects of HMGB1, it is difficult to interpret the role of single receptor types by only investigating such phosphorylation events, and we considered the analysis of cytokine levels as a more reliable method for this purpose.

In conclusion, we demonstrate that HMGB1 can induce production of several proinflammatory mediators from rat Mφ via phosphorylation of MAP kinases and translocation of NF-κB. We also demonstrate that HMGB1 exerts these proinflammatory effects by binding to RAGE, and that IL-1RI and TLR2 do not appear to play significant roles in signalling of HMGB1. Considering the significant influx of Mφ at the site of the local inflammation, their ability to produce a range of tissue-damaging mediators including NO, TNF and matrix metalloproteinases, and that HMGB1 is released by necrotic cells, we postulate that HMGB1 is a key mediator of chronic inflammation and will be an important therapeutic target in chronic diseases.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

This work was supported by grants from The Swedish Science Council, the Swedish Association against Rheumatism, B. von Kantzow's Foundation, King Gustaf V's Foundation, B. Dahlin's Foundation, U.&G. af Ugglas' Foundation, L.&H. Osterman's Foundation, the Freemason Lodge Barnhuset in Stockholm and Swedish Foundation for Strategic Research. The authors wish to acknowledge the help from Prof. Hans-Gustaf Ljunggren with accessing the RAGE-deficient mice.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References