HMGB1 is a potent trigger of arthritis

Transcription

1 Journal of Internal Medicine 2004; 255: MINISYMPOSIUM HMGB1 is a potent trigger of arthritis U. ANDERSSON 1,2 & H. ERLANDSSON-HARRIS 1 From the 1 Department of Medicine, Rheumatology Research Unit, Karolinska Institutet, Karolinska Hospital; and 2 Department of Woman and Child Health, Karolinska Institutet, Astrid Lindgren Children s Hospital, Stockholm, Sweden Abstract. Andersson U, Erlandsson-Harris H (Karolinska Institutet, Stockholm, Sweden). HMGB1 is a potent trigger of arthritis (Minisymposium). J Intern Med 2004; 255: Rheumatoid arthritis is a chronic inflammatory disease characterized by synovial inflammation and structural damage of joints. Although the cause of rheumatoid arthritis (RA) remains unknown, the excessive production of proinflammatory cytokines such as tumour necrosis factor (TNF) and interleukin-1 (IL-1) by intra-articular macrophages occupies a critical pathogenic role in the development and progression of the disease. High mobility group box chromosomal protein 1 (HMGB1) is a recently identified mediator of interest in human and experimental arthritides. HMGB1 can either be actively secreted from macrophages or passively released from necrotic cells of all kinds. Activated macrophages and unprogrammed cell death caused by ischaemia or activated complement are all prominent features of chronic arthritis, contributing to the persistent synovial inflammation. HMGB1 is cytoplasmically and extracellularly overexpressed in inflammatory synovial tissue in human RA as well as experimental collagen-induced arthritis. Elevated levels of HMGB1 are also present in synovial fluid samples from RA patients. Synovial tissue from rats with experimental arthritis exhibits aberrant deposition of HMGB1 preceding the onset of clinical signs of arthritis, and the expression becomes prominent after the onset of clinical disease. The synovial levels of HMGB1 are comparable with those of TNF and IL- 1b at the peak of manifest disease. HMGB1-targeted intervention with either neutralizing antibodies or the antagonistic A box domain of HMGB1 ameliorates collagen-induced arthritis both in mice and rats, and inhibits the local overexpression of IL-1b in the joints. It is thus conceivable that therapeutic HMGB1 blockade may contribute to future treatment of human chronic arthritis. Keywords: arthritis, cytokines, HMGB1, necrosis, 1 pathogenesis, therapy. Introduction Primary pathogenic events are unknown in most chronic arthritides, leaving control of inflammation as the next best option for treatment. The advent of selective tumour necrosis factor (TNF) and interleukin-1 (IL-1) blocking therapy illustrates the success of translational arthritis research based on characterization of cytokine networks operating in synovitis [1]. Approximately one-third of patients with rheumatoid arthritis (RA) show dramatically beneficial response to these novel therapies, whilst still a cohort of patients does not respond. The results warrant a search for modulation of additional inflammatory mediators to improve treatment for all patients with chronic arthritis. Therapeutic 344 Ó 2004 Blackwell Publishing Ltd

2 MINISYMPOSIUM: HMGB1 IN ARTHRITIS 345 neutralization of excessive, extracellular high mobility group box chromosomal protein 1 (HMGB1) may offer a new approach for managing various inflammatory diseases [2 5]. HMGB1 is a well-characterized DNA-binding protein [reviewed in 6, 7] that is also a potent proinflammatory molecule when released extracellularly [2, reviewed in 8, 9]. Indications that HMGB1 may act as a cytokine of great interest in rheumatology research is listed below: it is a potent inducer of macrophage activation including production of TNF, IL-1 and other proinflammatory mediators [2, 10, 11]; HMGB1 is actively secreted from stimulated macrophages/monocytes [2, 12]; it is passively released from necrotic cells of any kind; necrotic Hmgb1 )/) cells mediate minimal inflammatory response to injury [13, reviewed in 14]; extracellular HMGB1 is abundantly expressed in synovitis and intra-articular fluid of RA patients as well as in experimental arthritis [5, 11, 15]; aberrant synovial HMGB1 expression precedes the 2 clinical onset of collagen-induced arthritis (K. Palmblad, E. Sundberg, R. Kokkola, K.J. Tracey, U. Andersson, H. Erlandsson Harris, manuscript submitted); intra-articular HMGB1 injection induces synovitis in animal experiments [16]; cell membrane-expressed HMGB1 promotes local tissue invasion [17] and may have a causal connection to pannus-induced structural damage; HMGB1 can be successfully therapeutically targeted in a standard preclinical model of arthritis [5]. These novel aspects of HMGB1 biology will be the prime focus of this review. HMGB1 as a nuclear protein HMGB1 was isolated three decades ago as an abundant chromosomal protein with important structural functions in chromatin organization [18]. The molecule has a uniquely conserved sequence across species [6, 7]. The protein facilitates numerous nuclear functions, including transcription, replication and V(D)J recombination, as well as the action of p53, steroid hormone receptors, and Hox proteins [6, 19]. The phenotype of Hmgb1- deficient mice underscores the important role of HMGB1 as a regulator of transcription, as these animals die within a few hours from birth as a result of inefficient activation of glucocorticoid receptorresponsive genes [20]. Importantly, it has proved feasible to propagate cell lines from Hmgb1-deficient mice, although the gene abolition is not compatible with the postdelivery life of the animal. The first evidence that HMGB1 may also be expressed outside the cell nucleus came from studies of amphoterin, the gene of which was later shown to be identical to HMGB1 [17, 19]. Amphoterin was originally identified as a developmentally regulated protein abundantly expressed in the embryonic brain and in transformed cell lines [17, 21]. The protein is highly enriched in lamellopodia, the leading edges of migrating neuronal or transformed cells. HMGB1 plays a key role in regulating cell migration and metastasis [17, 21]. Local synthesis of HMGB1 increases its concentration in the growing protrusion and leads to its export into the extracellular space. The membrane-associated and the soluble extracellular protein will bind and signal via the receptor for advanced glycation end products (RAGE) [22 24]. This receptor belongs to the immunoglobulin superfamily and binds a variety of ligands: the glycated proteins that are present in the serum of diabetic patients (advanced glycation endproducts, AGE), calgranulin, and amyloid ß-peptide in Alzheimer patients [reviewed in 25]. It is expressed on a wide set of cells, including endothelial cells, smooth muscle cells, mononuclear phagocytes and neurones and has been implicated in several pathological processes, such as diabetes mellitus, neurodegenerative disorders, tumour invasion and atherosclerosis. HMGB1 binding on the cell surface itself induces the transcriptional upregulation of RAGE. The HMGB1 interaction with RAGE has several consequences: it signals to the cell motility system by activating Cdc42 and Rac (two guanosine triphosphatases that regulate cell motility) [23]; it signals to the metastatic pathway by recruiting plasminogen and its activator, thereby initiating a proteolytic process that facilitates tissue penetration [24, 26]; it signals to the nucleus by activating several mitogen-activated protein kinases leading to a nuclear translocation of NF-jB conveying an inflammatory response [23, 24]. HMGB1 as a proinflammatory cytokine The recognition of HMGB1 as a cytokine originates from studies of endotoxemia and sepsis [2]. The search for a downstream or late mediator of endotoxin lethality led to the unexpected discovery that HMGB1 is released, after a significant delay,

3 346 U. ANDERSSON & H. ERLANDSSON-HARRIS from macrophage cultures activated by exposure to endotoxin, tumour necrosis factor (TNF), or other proinflammatory stimuli [2, 9, 12]. In order to reveal putative cytokines produced as late or delayed mediators of inflammation, macrophage cultures were stimulated and proteins that appeared in the supernatant h later were isolated and characterized. The timing of this release is distinctly different from that of the early proinflammatory cytokines, such as IL-1 and TNF, which are produced in the first hours after exposure to endotoxin. Pulse chase studies revealed that most of the HMGB1 released in the first 24 h after exposure to endotoxin is derived from a preformed cellular protein pool (2). The preformed HMGB1 in macrophages is mainly localized in the cell nucleus. Activation of monocytes/macrophages leads to morphological and ultrastructural changes that result in a redistribution of nuclear HMGB1 into cytoplasmic organelles [12] (Fig. 1). HMGB1 lacks a signal sequence, but large amounts of HMGB-1 are released into the extracellular milieu by activated monocyte/macrophages via a nonclassical pathway involving regulated exocytosis of secretory lysosomes [12]. A similar route is used by IL-1b secreted from activated macrophages. In vivo studies confirmed a delayed appearance of extracellular HMGB1. Serum concentrations of HMGB1 increased significantly 8 32 h after in vivo administration of lipopolysaccharide (LPS) or TNF in mice [2]. Once released, HMGB1 is able to activate several other cells involved in the immune response or inflammatory reactions, and can act as a cytokine itself [2, 9 11]. HMGB1 potently stimulates monocytes to secrete a specific subset of proinflammatory cytokines, including TNF and IL-1. HMGB1 induces de novo cytokine synthesis, as cytokine mrna levels are increased after HMGB1 stimulation [10]. In comparison with the well-known inflammatory stimulus LPS, HMGB1 causes a delayed and biphasic release of TNF [9]. It appears, therefore, that HMGB1 is not only secreted by macrophages in response to proinflammatory stimuli, but also provokes a delayed response on its own; thereby, it prolongs and sustains inflammation. A central role for endogenous HMGB1 as a proinflammatory mediator was revealed in studies using anti-hmgb1 antibodies to prevent lethality from endotoxemia [2]. Anti- HMGB1 antibodies conferred significant protection against lethality even when the first dose of antibodies was administered after the early TNF and IL-1 responses had resolved. Anti-HMGB1 antibodies have proved to be beneficial when therapy is initiated prior to the onset of increased HMGB1 levels, providing a treatment window of h. Injection of toxic doses of HMGB1 in mice Fig. 1 Cultured, resting macrophages stained by immunofluorescence for intracellular HMGB1. The strongest green FITC signals generated by HMGB1 are located in the nucleus of each cell, with a weak contribution from cytoplasmic HMGB1 (a). The same cells counterstained in the nuclei by blue DAPI-generated fluorescence are demonstrated in (b). A substantial proportion of the green-labelled nuclear HMGB1 has been translocated to the cytoplasm of each cell after a co-culture period of 24 h with endotoxin, which activates the macrophages (c). The nuclei of the corresponding cells are demonstrated by blue DAPI staining (d).

4 MINISYMPOSIUM: HMGB1 IN ARTHRITIS 347 leads to the development of fever, weight loss, piloerection, shivering, and microthrombi formation in liver and lungs. Structure function analysis reveals that the B box of HMGB1 contains the proinflammatory cytokine domain. Recombinant B box, but not other recombinant truncated domains of the molecule, stimulates macrophages to release TNF and other proinflammatory cytokines from monocytes/macrophages and neutrophils [4]. In contrast, the A box is an anti-inflammatory molecule that competitively inhibits the proinflammatory activities of the full-length HMGB1 [27]. There is no evidence, yet, that the A box or B box are produced as distinct endogenous cytokines, but this possibility warrants further study because of the highly conserved nature and the newly revealed, distinctly antagonistic immunological activities of these protein domains. In addition to the regulated release of HMGB1 by activated monocytes/macrophages, preformed HMGB1 can also be released as a consequence of cell injury or unprogrammed death [13]. Most mammalian cells contain a large preformed nuclear pool of HMGB1 that is passively released during cellular necrosis. Lysis of normal, intact cells results in a protein mixture that can activate macrophages to release proinflammatory cytokines. The response reflects the known pro-inflammatory role of necrotic tissue. This concept may be central to the biology of HMGB1 as a cytokine, because it provides a mechanism that integrates the inflammatory response to infection and the inflammatory response to cell injury as occurs during ischaemia or complementmediated cell damage that occur in autoimmune diseases. Studies of necrotic or damaged Hmgb1 )/) cells in cocultures with macrophages reveal a greatly reduced ability to promote necrosis-induced inflammation, proving the importance of the release of HMGB1 as a proinflammatory signal from the necrotic cell [13]. Cell death by apoptosis is significantly less inflammatory, and cells dying from apoptosis do not release significant amounts of HMGB1. The molecular basis for decreased HMGB1 release during apoptosis is explained by the fact that HMGB1 remains tightly bound to the chromatin due to deacetylation of histone proteins. Together, this implicates a significant role for HMGB1 as a signal of accidental cell death that initiates immune response in the setting of cell necrosis, but not apoptosis. HMGB1 levels in disease Serum HMGB1 levels are below 5 ng ml )1 in the serum of healthy animals and normal human subjects when assayed using a quantitative immunoblot based on neutralizing anti-hmgb1 antibodies and a standard curve constructed with recombinant HMGB-1. Much higher circulating levels of HMGB1 (up to 150 ng ml )1 ) were observed in patients with severe sepsis, and the highest levels were assessed in patients who died [2]. Haemorrhagic shock is also associated with significantly increased serum levels of HMGB1, even in the absence of infection or endotoxemia [28]. This finding indicates that the stimulus to HMGB1 release Fig. 2 Micrographs illustrating immunostaining of HMGB1 (brown peroxidase) in erosive synovitis occurring in rodent collagen-induced arthritis (a) and in normal synovial tissue (b). The specimens are also counterstained with haematoxylin demonstrating blue nuclear staining. The pannus tissue invading the articular cartilage contains many cells with aberrant cytoplasmic HMGB1 expression (apart from the expected nuclear appearance) and additional extracellular HMGB1 (a). The noninflamed synovial membrane only displays a nuclear HMGB1 content (b). Certain cells do not reveal any HMGB1 content, probably due to low sensitivity of the staining methods.

5 348 U. ANDERSSON & H. ERLANDSSON-HARRIS may be either proinflammatory cytokines released by ischaemia or direct ischaemia-induced cell injury. HMGB1 and arthritis An aberrant extracellular or cytoplasmic HMGB1 expression has recently been demonstrated in human and experimental synovitis and implicates HMGB1 in the pathogenesis of arthritis [5, 11, 15, 16]. Immunohistochemical staining for HMGB1 in synovial tissue obtained from rodents with collagenor adjuvant-induced arthritis reveals that HMGB1 expression is significantly increased extracellularly and in the cytoplasm of macrophages and synoviocytes (Fig. 2a). This pattern differs significantly from the strictly nuclear HMGB1 staining pattern observed in synovial cells obtained from joints of normal mice and rats (Fig. 2b). The temporal and spatial synovial HMGB1 expression has recently been compared with that of TNF and IL-1b expression in longitudinal studies of collagen-induced arthritis (Palmblad K et al., unpublished data). The initial, expected distinct nuclear display of HMGB1 at early disease-preceding time points was subsequently followed by an unexpected inflammatory cytoplasmic HMGB1 expression within the synoviocytes of the nonproliferative lining layer, preceding clinical onset of disease by approximately 5 days. No corresponding increased TNF or IL-1 expression was recorded at this preclinical stage of disease. Pronounced cytoplasmic and additional extracellular HMGB1 expression coincided with the progression of clinical disease and was quantitatively comparable with that of TNF and IL-1b when maximal symptoms occurred. All three cytokines were particularly abundant in regions with tissue destruction, where the proliferating synovial tissue had invaded cartilage and bone (Fig. 2a). The prominent HMGB1 pannus expression may be a primary event due to tissue ischaemia and activated complement or a secondary result caused by pro-inflammatory stimuli such as TNF or IL-1. The pronounced HMGB1 expression in the pannus tissue may be involved in the destructive cartilage and bone penetrating process. In analogy, HMGB1 is important for tumour invasion, a process mediated via RAGE interaction [24, 26]. Additional support for a pathogenic role of HMGB1 for the development of arthritis comes from studies of intra-articular HMGB1 injections in mouse knee joints [16]. A single injection of rhmgb1 resulted in mild to moderate synovitis that persisted for at least 28 days. The incidence and severity of articular inflammation varied between different mouse strains. The majority of cells found in the inflamed synovial membrane were macrophages with only a few detectable T lymphocytes. No significant differences were found with respect to incidence and severity of arthritis between mice depleted of monocytes, granulocytes, or lacking T/B lymphocytes. However, combined removal of monocytes and neutrophils resulted in a significantly lower incidence of arthritis. Mice rendered deficient in the IL-1 receptor did not develop inflammation after a single intra-articular HMGB1 injection. Taken together, these results indicate that HMGB-1 is not a mere expression of inflammatory responses, but on its own triggers joint inflammation. We recently examined clinical effects in collageninduced arthritis using therapeutic administration of neutralizing HMGB1 antigen-affinity purified polyclonal antibodies or truncated HMGB1-derived A-box protein, a specific, competitive antagonist of HMGB1 [5]. The effects of HMGB1-blocking therapy were studied on established disease, as this may be more relevant to the design of clinical trials than a pretreatment approach. Systemic administration of anti-hmgb-1 antibodies or A-box protein did not mediate detectable toxic effects and significantly reduced the mean arthritis score, the diseaseinduced weight loss, and the histological severity of arthritis. Beneficial effects were observed both in mice and rats. Immunohistochemical analysis revealed pronounced synovial IL-1b expression and articular cartilage destruction in vehicle-treated mice. Both these features were significantly less manifested in animals treated with anti-hmgb1 antibodies or A-box protein. The beneficial results of the present protocols were comparable with those obtained in previous studies using anti-tnf antibody treatment of established murine collageninduced arthritis. HMGB1- and TNF-blockade regimens both led to clinical improvement within 1 2 days. Future studies are required to examine whether therapeutic targeting of HMGB1 or TNF may provide selective opportunities or drawbacks in treatment of arthritis. Presently it is not fully understood to what extent HMGB1 depends on TNF to mediate its proinflammatory function. Anal-

6 MINISYMPOSIUM: HMGB1 IN ARTHRITIS 349 ogous treatment based on neutralizing anti-hmgb1 antibodies or truncated A-box protein confers significant clinical protection in other experimental inflammatory conditions such as sepsis [27], endotoxemia [2] and lung inflammation [3]. Evidence for a role for HMGB1 in the pathogenesis of arthritis is not only based on observations in preclinical models. Two separate research groups have recently reported about aberrant, intra-articular of HMGB1 in RA, detectable in the synovial tissue as well as in the fluid [11, 15]. HMGB1 synovial fluid levels are significantly elevated in patients with RA compared with those from joints with osteoarthritis. Synovial fluid macrophages exhibit increased expression of RAGE and can be activated to release TNF, IL-1b, and IL-6 by exposure to HMGB1. Exposure of macrophages from the RA synovium to TNF activated a translocation of HMGB1 from the nucleus to the cytosol. Blockade by srage inhibited the release of TNF from synovial fluid mononuclear cells [11]. Thus, synovial monocyte/ macrophages in arthritic joints can respond to inflammatory agents releasing HMGB1 and can be activated by HMGB1 to release proinflammatory cytokines. In accordance with findings of our research group [15], Taniguchi et al. reported that HMGB1 is localized primarily in the cytosol of macrophage-like cells located in the sublining layer of the synovial membrane. The cells exhibited the characteristics and expected distribution of synovial macrophages, confirmed by the fact that cells with cytosolic HMGB1 expression coexpressed macrophage surface markers. HMGB1 release from synovial fluid macrophages was enhanced by exposure of the cultures to TNF. The aberrant HMGB1 expression in the inflamed synovial tissue is particularly prominent in vascular high endothelial cells and in certain synoviocytes in the lining layer and in lymphoid aggregates. The functional relevance of these descriptive findings needs to be resolved. Divergent results have been reported about systemic levels of HMGB1 in RA patients. Taniguchi et al. were unable to detect increased serum levels of HMGB1, whilst Tracey et al. demonstrated detectable serum HMGB1 in all studied patients with active RA [29]. No correlation was seen in this small study with regard to disease activity and serum HMGB1 levels. It is likely that methods and antibody reagents have an important impact on the differences recorded in the two studies. Considered together, these data indicate that HMGB1 can contribute to the pathogenesis of inflammatory diseases, including arthritis. Prominent features of chronic synovitis are necrotic cell death and activation of macrophages. Both of these processes release HMGB1, which can contribute to the inflammation of chronic arthritis. HMGB1 stimulates synovial macrophages to produce and release TNF, IL-1b and IL-6; it can bind to components of the plasminogen activation system and enhance the activity of tissue plasminogen activator and of matrix metalloproteinases (MMP), such as MMP-2 and MMP-9. It is thus plausible that HMGB1 plays a role in both the inflammatory and destructive processes in the pathophysiology of arthritis. The clinical results of the HMGB1 blockade therapies in collagen-induced arthritis are promising and are of great conceptual interest. From a practical point of view, it is conceivable that refinement of HMGB1 antagonists should further improve these results. The development of monoclonal antibodies against HMGB1 or the A box peptide fused to a larger molecule to retard elimination may possibly provide therapeutic benefit to arthritis patients. Studies on rheumatology have benefited to a great extent from the original sepsis research work on TNF biology that subsequently led to beneficial TNF blocking intervention in arthritis [30]. There is a possibility that history will repeat itself with HMGB1, discovered as a mediator in sepsis, to represent a novel therapeutic target molecule in chronic arthritis. Conflict of interest statement Our research group has received a 2-year grant from Critical Therapeutics Inc., a commercial company in Boston, MA, developing antibodies against HMGB1. References 1 Taylor PC, Williams RO, Maini RN. Immunotherapy for rheumatoid arthritis. Curr Opin Immunol 2001; 13: Wang H, Bloom O, Zhang M et al. HMG-1 as a late mediator of endotoxin lethality in mice. Science 1999; 285: Abraham E, Arcaroli J, Carmody A, Wang H, Tracey KJ. Cutting edge: HMG-1 as a mediator of acute lung inflammation. J Immunol 2000; 165: Li J, Kokkola R, Tabibzadeh S et al. Structural basis for the proinflammatory cytokine activity of high mobility group box 1 (HMGB1). Mol Med 2003; 9: Kokkola R, Sundberg E, Ulfgren A-K et al. Successful treatment of collagen-induced arthritis in mice and rats by

7 350 U. ANDERSSON & H. ERLANDSSON-HARRIS targeting extracellular high mobility group box chromosomal protein 1 activity. Arthritis Rheum 2003; 48: Bustin M. Regulation of DNA-dependent activities by the functional motifs of the high mobility group chromosomal proteins. Mol Cell Biol 1999; 19: Bianchi ME, Beltrame M. Upwardly mobile proteins. Workshop: the role of HMG proteins in chromatin structure, gene expression and neoplasia. EMBO Rep 2000; 1: Andersson U, Erlandsson-Harris H, Yang H, Tracey KJ. HMGB1 as a DNA-binding cytokine. J Leukoc Biol 2002; 72: Yang H, Wang H, Tracey KJ. HMG-1 rediscovered as a cytokine. Shock 2001; 15: Andersson U, Wang H, Palmblad K et al. High mobility group 1 protein (HMG-1) stimulates proinflammatory cytokine synthesis in human monocytes. J Exp Med 2000; 192: Taniguchi N, Kawahara K, Yone K et al. High mobility group box chromosomal protein 1 plays a role in the pathogenesis of rheumatoid arthritis as a novel cytokine. Arthritis Rheum 2003; 48: Gardella S, Andrei C, Ferrera D et al. The nuclear protein HMGB1 is secreted by monocytes via a non-classical, vesiclemediated secretory pathway. EMBO Rep 2002; 3: Scaffidi P, Misteli T, Bianchi ME. Release of chromatin protein HMGB1 by necrotic cells triggers inflammation. Nature 2002; 418: Bustin M. At the crossroad of necrosis and apoptosis: signaling to multiple cellular targets by HMGB1. Sci STKE 2002; 3 Sep 24: 151: pe Kokkola R, Sundberg E, Ulfgren AK et al. High mobility group box chromosomal protein 1: a novel proinflammatory mediator in synovitis. Arthritis Rheum 2002; 46: Pullerits R, Jonsson I-M, Verdrengh M et al. High mobility group box chromosomal protein 1, a DNA binding cytokine, induces arthritis. Arthritis Rheum 2003; 48: Merenmies J, Pihlaskari R, Laitinen J, Wartiovaara J, Rauvala H. 30-kDa heparin-binding protein of brain (amphoterin) involved in neurite outgrowth: amino acid sequence and localization in the filopodia of the advancing plasma membrane. J Biol Chem 1991; 266: Goodwin GH, Johns EW. Isolation and characterisation of two calf-thymus chromatin non-histone proteins with high contents of acidic and basic amino acids. Eur J Biochem 1973; 40: Bustin M. Revised nomenclature for high mobility group (HMG) chromosomal proteins. Trends Biochem Sci 2001; 26: Calogero S, Grassi F, Aguzzi A et al. The lack of chromosomal protein Hmg1 does not disrupt cell growth but causes lethal hypoglycaemia in newborn mice. Nat Genet 1999; 22: Sajithlal G, Huttunen H, Rauvala H, Munch G. Receptor for advanced glycation end products plays a more important role in cellular survival than in neurite outgrowth during retinoic acid-induced differentiation of neuroblastoma cells. J Biol Chem 2002; 277: Hori O, Brett J, Slattery T et al. The receptor for advanced glycation end products (RAGE) is a cellular binding site for amphoterin: mediation of neurite outgrowth and co-expression of RAGE and amphoterin in the developing nervous system. J Biol Chem 1995; 270: Huttunen HJ, Fages C, Rauvala H. Receptor for advanced glycation end products (RAGE)-mediated neurite outgrowth and activation of NF-jB require the cytoplasmic domain of the receptor but different downstream signaling pathways. J Biol Chem 1999; 274: Taguchi A, Blood DC, del Toro G et al. Blockade of RAGEamphoterin signalling suppresses tumour growth and metastases. Nature 2000; 405: Schmidt AM, Yan SD, Yan SF, Stern DM. The biology of the receptor for advanced glycation end products and its ligands. Biochim Biophys Acta 2000; 1498: Parkkinen J, Rauvala H. Interactions of plasminogen and tissue plasminogen activator (t-pa) with amphoterin: enhancement of t-pa-catalyzed plasminogen activation by amphoterin. J Biol Chem 1991; 266: Yang H, Wang H, Li J et al. Reversing established sepsis with antagonists of endogenous HMGB1. Proc Natl Acad Sci, USA 2004; 101: Ombrellino M, Wang H, Ajemian MS et al. Increased serum concentrations of high-mobility-group protein 1 in haemorrhagic shock. Lancet 1999; 354: Ulloa L, Batliwalla FM, Andersson U, Gregersen PK, Tracey KJ. HMGB1 as a nuclear protein, cytokine and potential therapeutic target in arthritis. Arthritis Rheum 2003; 48: Tracey KJ, Fong Y, Hesse DG et al. Anti-cachectin/TNF monoclonal antibodies prevent septic shock during lethal bacteraemia. Nature 1987; 330: Correspondence: Professor Ulf Andersson, Department of Rheumatology, Astrid Lindgren Children s Hospital, Q1:02,171 76, Stockholm, Sweden. (fax: ; ulf@mbox313.swipnet.se)