Reinhard E. Voll, Department Rheumatology and Clinical Immunology & Centre for Chronic Immunodeficiency, University Medical Centre Freiburg, Hugstetter Str. 55, 79106, Freiburg/Breisgau, Germany. (fax: +49761 270 34460; e-mail: firstname.lastname@example.org).
Abstract. Urbonaviciute V, Voll RE (Friedrich-Alexander University of Erlangen-Nuremberg, Erlangen; and University Medical Centre Freiburg, Freiburg/Breisgau; Germany). High-mobility group box 1 represents a potential marker of disease activity and novel therapeutic target in systemic lupus erythematosus (Symposium). J Intern Med 2011; 270: 309–318.
High-mobility group box 1 (HMGB1) protein is a nuclear DNA-binding protein, which functions as an alarmin when released from cells. Recent studies implicate extracellular HMGB1 in the pathogenesis of systemic lupus erythematosus (SLE), a prototypical autoimmune disease characterized by the formation of multiple autoantibodies, especially those directed against nucleosomes and double-stranded (ds)DNA. Elevated concentrations of HMGB1 are observed in sera as well as in skin lesions of patients with lupus. Of importance, serum HMGB1 and anti-HMGB1 autoantibody levels correlate with disease activity. In the blood of patients with SLE, HMGB1 is complexed with nucleosomes, at least partially. Moreover, HMGB1–nucleosome complexes from apoptotic cells activate antigen-presenting cells. Injection of HMGB1–nucleosome complexes into nonautoimmune mice results in the formation of autoantibodies against dsDNA and histones in a Toll-like receptor (TLR) 2-dependent manner. Additionally, HMGB1, as a part of DNA–anti-DNA immune complexes, can interact with receptor for advanced glycation end products (RAGE) on the surface of plasmacytoid dendritic cells and B cells leading to TLR9-dependent interferon (IFN)α release and activation of autoreactive B cells, respectively. HMGB1 attached to neutrophil extracellular traps may contribute to IFNα production by facilitating the recognition of self-nucleic acids. Furthermore, HMGB1, complexed with DNA and pathogenic anti-DNA autoantibodies, activates its receptors, TLR2, TLR4 and RAGE, and may thereby be involved in anti-DNA autoantibody-induced kidney damage in lupus nephritis. Collectively, these findings suggest that HMGB1 is a potential marker of disease activity and, because of its probable involvement in the pathogenesis, a novel therapeutic target in SLE.
High-mobility group box 1 protein in the cell nucleus
High-mobility group box 1 (HMGB1) is a ubiquitously expressed, abundant architectural chromosomal protein of 215 amino acids with a highly conserved sequence across species . Mammals express two additional HMG proteins, HMGB2 and HMGB3, which appear to be structurally and functionally very similar to HMGB1 but have a restricted expression pattern. HMGB1 consists of three domains, two basic DNA-binding motifs termed HMGB domains A and B, and a highly acidic C-terminal domain, which together confer a bipolar charge to the protein . In the cell nucleus, HMGB1 binds without sequence specificity to the minor groove of DNA and induces important modifications in the DNA structure, particularly strong bends. In addition, HMGB1 recognizes and binds to highly structured noncanonical or damaged DNA such as four-way junctions, hemicatenanes, minicircles and ultraviolet light (UV)-irradiated DNA [3, 4]. These properties of HMGB1 enable it to participate in many DNA-related biological processes, including transcription, chromatin remodelling, DNA damage repair and recombination. In addition to DNA bending, HMGB1 facilitates the formation of multiple nucleoprotein complexes by protein–protein interactions. HMGB1 binds to different transcription factors, DNA repair proteins, site-specific recombination proteins, histones and other molecules. HMGB1 is also involved in the regulation of the nucleosomal structure via binding to the linker DNA and to histones of the nucleosome core particle. Post-translational modifications such as acetylation of HMGB1 and/or core histones affect the affinity of HMGB1 to nucleosomes .
The biological relevance of HMGB1 in vivo is shown in HMGB1-deficient mice. HMGB1−/− mice are born with several defects, such as open eyelids, lung atelectasis and lethargy, and die soon after birth because of severe hypoglycaemia caused by deficient glucocorticoid receptor function. Cell lines obtained from these animals grow normally, but activation of gene expression by some transcription factors, including glucocorticoid receptors, is impaired .
Extracellular HMGB1 as an endogenous alarmin
Wang et al.  first described extracellular HMGB1 as a late-acting proinflammatory mediator of endotoxaemia. Following stimulation with inflammatory mediators including tumour necrosis factor (TNF) α and Toll-like receptor (TLR) ligands such as lipopolysaccharides (LPS)  or polyinosinic–polycytidylic acid , HMGB1 is actively secreted through a nonclassical Golgi-independent mechanism by monocytes/macrophages, dendritic cell (DC), natural killer cells and others [9, 10].
Scaffidi et al.  reported an alternative mechanism for HMGB1 release: HMGB1 normally is loosely bound to chromatin, and bioactive HMGB1 diffuses into the extracellular space if the cells become necrotic. By contrast, HMGB1 is not released during the early stages of apoptotic cell death. Hypoacetylation of chromosomal proteins, phosphorylation of histone H2B and, possibly, other as yet unidentified modifications during apoptotic cell death lead to tight binding of HMGB1 to the chromatin, preventing its release and avoiding inflammation. The release of HMGB1 upon necrotic but not early apoptotic cell death may crucially contribute to the inflammatory response, which is characteristic of necrosis. However, recent findings demonstrated that HMGB1 release may not be an exclusive feature of primary necrosis. HMGB1 release also occurs late in the apoptotic cell death process, at a stage known as secondary necrosis , in which HMGB1 is partially released as a complex with chromatin . It is interesting that DNA fragmentation catalysed by caspase-activated DNase or DNase γ is a prerequisite for HMGB1 release following apoptotic stimuli in certain cell types . This is consistent with our observation of HMGB1 mainly complexed with mono- and oligonucleosomes in the supernatants from secondary necrotic cells .
When released from primary or secondary necrotic cells, HMGB1 behaves as a typical alarmin. The term alarmin was proposed by Oppenheim and Yang  to describe structurally diverse endogenous molecules that signal cell and tissue damage, also known as endogenous danger signals. Alarmins have several characteristics: (i) they are rapidly released from damaged cells, (ii) can be produced and released by some activated cells without causing the death of these cells, (iii) they recruit and activate immune cells, thereby augmenting innate and adaptive immune responses, and (iv) they promote repair of damaged tissues [15, 16].
At least three receptors have been reported to mediate the proinflammatory and immune-activating effects of extracellular HMGB1: (i) receptor for advanced glycation end products (RAGE), (ii) TLR2, and (iii) TLR4. Stimulation of these receptors leads to activation of the transcription factor NF-κB, inducing the transcription of multiple pro-inflammatory genes . Upon (co-)activation with HMGB1, macrophages produce pro-inflammatory cytokines such as TNFα, interleukin (IL)-1β, IL-6, IL-8, macrophage inflammatory protein (MIP) 1α and MIP2β. In addition, HMGB1 induces migration and maturation of DCs, which is characterized by expression of HLA-DR, CD83, CD80 and CD86 [18, 19]. Furthermore, HMGB1 regulates T-cell activation .
Recent studies indicate that complex formation of HMGB1 with other structurally different molecules can increase the biological activity of both components. HMGB1 acts as a cofactor for several TLR ligands. Tian et al. demonstrated that HMGB1 in complex with a TLR9 ligand, type A CpG (CpG-A) oligodeoxynucleotides (ODN), binds more efficiently to RAGE than HMGB1 alone. The CpG-A ODN-mediated release of interferon (IFN)α from total bone marrow cells and plasmacytoid DCs (pDCs) was strongly enhanced by HMGB1 and was dependent on the presence of RAGE and TLR9 . Moreover, Ivanov et al. demonstrated that macrophages and DCs secrete HMGB1 upon treatment with CpG ODN. Extracellular HMGB1 in turn binds to CpG ODN and enhances CpG ODN-triggered cytokine responses, possibly by accelerating and increasing the recognition of CpG ODN by TLR9. Moreover, HMGB1 interacts and pre-associates with TLR9 even prior to ODN treatment in the endoplasmic reticulum–Golgi intermediate compartment, and hastens the redistribution of TLR9 to early endosomes in response to CpG-ODN . In addition, HMGB1 binds LPS in a concentration-dependent manner and transfers it to soluble and membrane-bound CD14, which concentrates LPS and increases the sensitivity of LPS signalling via the TLR4–MD2 pathway . The binding of HMGB1 to the TLR1/TLR2 ligand Pam3CSK4 augments production of proinflammatory cytokines by peripheral blood mononuclear cells . Furthermore, HMGB1 acts synergistically when complexed with cytokines such as IL-1β and markedly enhances their proinflammatory activity. Anti-IL-1β antibodies and an IL-1 receptor (IL-1R) antagonist block the proinflammatory activity of the HMGB1/IL-1β complex, indicating the involvement of the IL-1R . Recently, it was reported that HMGB proteins represent universal sentinels for the recognition of virtually all immunogenic nucleic acids, especially of viral and microbial origin. HMGB proteins are critical for triggering TLR3-, TLR7- and TLR9-mediated innate immune responses by their cognate nucleic acids .
High-mobility group box 1 undergoes various post-translational modifications, including oxidation, methylation, glycosylation, phosphorylation and acetylation . We showed recently that during apoptotic and necrotic cell death, HMGB1 undergoes reversible oxidative modifications of cysteine residues, which may modulate its biological properties, including interactions with other proteins and DNA. Additionally, oxidation of HMGB1 may lead to generation of neoepitopes prone to promote autoimmune responses to HMGB1 as detected in many autoimmune diseases, but also in some healthy individuals . Kazama et al.  demonstrated that HMGB1 that had been oxidized during apoptosis at cysteine residue 106 loses its direct proinflammatory properties. Moreover, Yang et al.  reported that the cysteine residue at position 106 of HMGB1 is required for the interaction of HMGB1 with TLR4, and induction of TNFα secretion by macrophages. Yet the processes leading to oxidative modifications of HMGB1 and their functional consequences remain incompletely understood.
Recent studies also implicate HMGB1 in the negative regulation of inflammation. HMGB1 and other endogenous danger signals such as heat-shock proteins (HSPs), but not molecules of microbial origin, interact with CD24, a glycosylated glycosyl phosphatidylinositol-anchored membrane protein. CD24, in turn, binds sialic acid-binding immunoglobulin (Ig)-like lectin (Siglec) 10 in humans and Siglec G in mice; both these Siglecs are considered to be negative regulators of the immune response. Hence, the HMGB1–CD24–Siglec G pathway might protect the host against a lethal response to pathological cell death and discriminate between damage- and pathogen-associated molecular patterns (DAMPs and PAMPs) .
Another negative regulator of HMGB1 activity is thrombomodulin, an integral membrane protein expressed on the surface of vascular endothelial cells that functions as a cofactor in the thrombin-induced activation of protein C in the anticoagulant pathway. The N-terminal lectin-binding domain of thrombomodulin binds HMGB1 and blocks its interaction with RAGE .
The alarmin HMGB1 in the aetiopathogenesis of lupus
Systemic lupus erythematosus (SLE) is a typical autoimmune disease that is predominantly observed in women of childbearing age. Genetic and environmental factors such as smoking, UV exposure and infections play a role in disease development. Patients with SLE can present a broad range of clinical manifestations including skin rash, photosensitivity, arthritis, glomerulonephritis, neurological symptoms, haemolytic anaemia and vasculitis. Organ damage occurs as a consequence of the chronic, uncontrolled autoimmune response directed against ubiquitous, mostly nuclear, self-antigens, or secondarily because of ischaemia caused by vasculitis or thrombosis. Nevertheless, to date, the aetiology and pathogenetic mechanisms of SLE have not been fully elucidated .
There is increasing evidence that HMGB1 contributes to the pathogenesis of lupus as a result of its proinflammatory and immunostimulatory properties. In cutaneous lupus, increased amounts of cytoplasmic and extracellular HMGB1 were detected within the lesional skin together with high expression levels of TNFα and IL-1β . A follow-up investigation revealed cytoplasmic and extracellular HMGB1 at the peak of clinical activity in experimentally UV-induced lesions of cutaneous lupus . Together, these findings suggest a role of HMGB1 in the local pro-inflammatory process and tissue damage in cutaneous lupus.
Others and we have found elevated concentrations of systemically released HMGB1 in serum and plasma samples from patients with SLE and experimental lupus mouse models [13, 34–37]. Of note, serum HMGB1 concentrations positively correlate with the SLE Disease Activity Index (SLEDAI) and inversely correlate with the levels of complement components C4 and C3, whereby decreased concentrations of these components may indicate complement consumption because of active disease. Moreover, concentrations of HMGB1 in serum were higher in patients with vasculitis and myositis as well as in patients with renal involvement [36, 38]. Therefore, HMGB1 may represent a potential marker of disease activity. However, these studies have not been able to determine the origin of HMGB1 or the mechanisms whereby HMGB1 contributes to the aetiopathogenesis of SLE.
Impaired phagocytosis of apoptotic or necrotic cells with consecutive release of nuclear autoantigens/potential endogenous danger signals may predispose to lupus and disease exacerbation [39, 40]. Recognition and uptake of apoptotic cells by phagocytes, known as efferocytosis, is a complex process that involves a wide variety of markers on the apoptotic cells, multiple phagocytic receptors and soluble molecules that opsonize apoptotic cells and create molecular bridges between components on the apoptotic cells and phagocyte surfaces [41, 42]. It was recently reported that extracellular HMGB1 could reduce the elimination of dying cells by interacting with several components of the apoptotic cell clearance system. HMGB1 associates with phosphatidylserine (PS) on the outer leaflet of the cell membrane of apoptotic cells, which prevents its activity as an ‘eat me’ signal for phagocytes . On the surface of phagocytes, HMGB1 binds to integrin αvβ3, a receptor involved in PS recognition, and thereby suppresses the interaction between αvβ3 and the opsonin milk fat globule EGF factor 8 (MFG-E8), which acts as a bridge between PS and αvβ3 [44, 45]. Moreover, it was shown that RAGE plays a role in efferocytosis, and the acidic C-terminal domain of HMGB1 is responsible for the inhibitory effects of HMGB1 on RAGE-mediated clearance of apoptotic cells in vitro and in vivo [44, 46]. Increased cell death as well as defects in the clearance of apoptotic cells may lead to an excess of apoptotic cells, overload of the clearance mechanisms and, consequently, secondary necrosis with the release of potential nuclear autoantigens. Hence, apoptotic especially secondary necrotic cells may foster the development of lupus, at least in a subgroup of patients [39, 47–49]. Specifically, in the case of delayed clearance of apoptotic cells, secondary necrosis can ensue, resulting in the exposure of potential danger signals and self-antigens that are usually buried within the cell . During apoptotic cell death, the endogenous alarmin HMGB1 becomes tightly bound to the chromatin . We hypothesized that HMGB1 may be released in complex with nucleosomes/chromatin from noneliminated apoptotic cells and promote inflammation and activation of antigen-presenting cells, thereby contributing to the aetiopathogenesis of SLE. We found that HMGB1 in fact remains bound to the nucleosomes liberated from secondary necrotic cells in vitro and to nucleosomes circulating in the blood of patients with SLE, which are usually derived from apoptotic cells. Also, immune complexes isolated from some patients with SLE contained HMGB1. It is noteworthy that HMGB1 in complex with nucleosomes either purified from apoptotic cells or released from secondary necrotic cells activated macrophages and DCs, which caused inflammatory cytokine secretion and enhanced antigen presentation (Fig. 1). Moreover, in vitro‘apoptotic’ nucleosome-induced cytokine release by macrophages was dependent on the presence of MyD88 and TLR2, whereas RAGE, TLR4 and TLR9 were not essential. It remains unclear whether the nucleosome or the HMGB1 molecule, or the complex of both, represents the TLR2-binding partner. Of importance, HMGB1-containing nucleosomes from apoptotic cells induced anti-dsDNA and anti-histone IgG responses in a TLR2-dependent manner in nonautoimmune mice (Fig. 2). The nucleosomes derived from live control cells were immunologically silent . It appears that RAGE is not required for this effect, as HMGB1-containing nucleosomes also elicited anti-dsDNA autoantibodies in RAGE-deficient mice. The importance of the TLR2 pathway in lupus development was confirmed in a mouse model of SLE induced by the mineral oil pristane. During infection, TLR2 in combination with TLR1 recognizes triacylated lipopeptides and in combination with TLR6 recognizes diacylated lipopeptides of Gram-positive bacteria. During cell damage, TLR2 binds molecules derived from dying and dead cells such as HSPs, HMGB1 and probably HMGB1–nucleosome complexes. The recognition of these endogenous ligands leads to activation of proinflammatory genes and may promote autoimmunity . We found that anti-dsDNA, anti-histone, anti-nucleosome and IgG responses to certain extractable nuclear antigens, characteristic of lupus or other connective tissue diseases, were significantly reduced in pristane-treated TLR2−/− mice compared to pristane-treated wild-type controls. Moreover, TLR2-deficient mice treated with the mineral oil were at least partially protected from lupus-like kidney disease (Urbonaviciute V, Starke C, Graef D, Pirschel W, Frey S, Schett G, Voll RE. 2011, in preparation). A role of TLR2 and TLR4 has also been suggested in the pathogenesis of SLE. The presence of a single loss-of-function allele of MAL/TIRAP, an adaptor protein which is important for TLR2 and TLR4 signalling, was found to be protective against SLE but had no influence on the occurrence of rheumatoid arthritis . In conclusion, HMGB1–nucleosome complexes activate antigen-presenting cells predominantly through TLR2 and may crucially contribute to the aetiopathogenesis of SLE via facilitating the breaking of immunological tolerance against chromatin (Fig. 3).
After breaking T and B cell tolerance, high-affinity IgG antibodies to dsDNA/nucleosomes can be produced. These autoantibodies may form immune complexes with dsDNA both in the tissues and in the circulation of patients with SLE. In fact, we detected HMGB1 within polyethylene glycol-precipitated immune complexes as well as in complex with nucleosomes in the blood of patients with SLE . Moreover, Tian et al.  showed that IFNα release from pDCs and expansion of autoreactive B cells by HMGB1–DNA-containing immune complexes were dependent on binding of HMGB1 to RAGE and activation of the TLR9–MyD88 pathway by DNA.
The results of a very recent study revealed that neutrophils from paediatric patients with lupus undergo accelerated cell death in culture . Furthermore, anti-ribonucleoprotein autoantibodies, detectable in sera of approximately one-third of patients with SLE , switch SLE neutrophil death from apoptosis to NETosis . NETosis is a unique form of death characterized by nuclear disruption and release of neutrophil extracellular traps (NETs), web-like structures that are highly enriched in chromatin components such as DNA, histones and HMGB1 [52, 54, 55]. Of note, these SLE NETs induce secretion of large amounts of IFNα by pDCs in a DNA- and TLR9-dependent manner . In this situation, HMGB1 complexed with self-DNA may act as a sentinel that facilitates the uptake and recognition of DNA or RNA by cytosolic nucleic acid-sensing receptors . Together, these findings demonstrate that HMGB1 may contribute to IFNα production, which plays an important role in the immunopathogenesis of SLE.
Lupus nephritis is a frequent complication in patients with SLE and is the major cause of death in most mouse lupus models. Glomerular deposition of pathogenic autoantibodies in the kidney, leading to in situ immune complex formation or deposition of preformed immune complexes, leads to complement activation as well as ligation of Fc receptors and eventually kidney damage in lupus. Qing et al. suggested that HMGB1 may act as a proinflammatory mediator in antibody-induced kidney damage in SLE. The pathogenic anti-DNA antibody 1A3F can bind to HMGB1–DNA complexes as well as to DNA-free purified HMGB1. Anti-DNA antibody and HMGB1, probably by activating its receptors TLR2 and RAGE, exhibited a synergistic effect on upregulation of proinflammatory gene expression in renal mesangial cells in vitro and chemokine expression in kidneys of BALB/c mice .
HMGB1 as an autoantigen in SLE
High-mobility group box 1 can be a target of autoimmune response: anti-HMGB1 antibodies were found in patients with several autoimmune diseases including rheumatoid arthritis and SLE [34, 38, 57–59]. Moreover, the presence of anti-HMGB1 antibodies positively correlates with SLEDAI and serum HMGB1 concentrations, and negatively with platelet number in patients with SLE, suggesting a pathological role of these antibodies. Epitope mapping revealed multiple HMGB1 epitopes recognized by antibodies present in serum from SLE patients, with the major epitopes mapping to box A [38, 60]. It is interesting that we also detected autoantibodies to HMGB1 in most healthy individuals. These autoantibodies were predominantly of the IgM isotype, but low titres of IgG autoantibodies were also frequently detectable. The presence of anti-HMGB1 antibodies in most healthy subjects might be because of cross-reactivity or the sticky nature of HMGB1. The HMGB1-binding antibodies, predominantly of the IgM isotype, might play an important physiological role by modulating the proinflammatory activity of HMGB1, thereby limiting overwhelming inflammatory responses caused by massive HMGB1 release in conditions such as sepsis or extensive necrosis. Of note, anti-HMGB1 autoantibodies as well as other HMGB1-binding serum proteins impede the reliable quantification of HMGB1 by enzyme-linked immunosorbent assay (ELISA). We reported previously that anti-HMGB1 autoantibodies, which are present at low levels in healthy individuals and often at high levels in patients with lupus, could interfere with the detection of HMGB1 by ELISA. Frequently, lower HMGB1 concentrations were detected by ELISA in comparison with Western blot analysis . Hence, semiquantitative immunoblot analysis appears to be more appropriate for the measurement of total HMGB1 concentrations in serum and plasma in patients with SLE. However, immunoblot analyses of HMGB1 in total serum or plasma are labour intensive and technically challenging because of background problems and therefore are not suitable for routine testing. Possibly, ELISA detects predominantly free, ‘bioavailable’ HMGB1, but this has not been investigated in detail yet. In addition, post-translational modifications as well as binding partners may critically modulate the biological activities of HMGB1. Therefore, reliable quantification of HMGB1 remains a problem, and results obtained from serum samples should be interpreted with care. Routine diagnostic methods to reliably quantify HMGB1 and to distinguish its functionally different modifications and complexes in serum and plasma are required to more specifically elucidate the role of this multifunctional protein as a diagnostic and/or prognostic marker and as a potential therapeutic target in immune and inflammatory diseases.
Systemic lupus erythematosus is a systemic autoimmune disease characterized by the breakdown of tolerance against self-antigens and subsequent damage to multiple tissues and organs. Nevertheless, the precise aetiopathogenesis of the disorder remains unclear.
Features of SLE include intense immune activation as well as increased apoptosis and/or a reduced clearance of dying cells; all these processes lead to HMGB1 release. Several studies, in fact, have shown elevated levels of locally and systemically released HMGB1 in experimental mouse models of lupus as well as in patients with SLE. Moreover, elevated HMGB1 concentrations positively correlated with disease activity.
Different mechanisms through which HMGB1 can act at different stages of disease development have been described. Extracellular HMGB1 inhibits elimination of apoptotic material, an abundant source of endogenous danger signals, including HMGB1 itself. Moreover, HMGB1–nucleosome complexes released from noneliminated secondary necrotic cells activate antigen-presenting cells and induce lupus-specific autoantibody formation in a TLR2-dependent manner, thereby probably contributing to the initiation of the autoimmune process in lupus. Furthermore, HMGB1–DNA–autoantibody complexes can activate the RAGE and TLR9 pathways and may then act as an amplification loop of autoimmunity and inflammation. Additionally, HMGB1 may mediate local renal damage through binding to pathogenic anti-DNA autoantibodies in lupus nephritis. Finally, anti-HMGB1 IgG antibodies have been found in patients with SLE; however, the clinical relevance of these autoantibodies remains to be determined.
Taken together, these findings suggest a role for HMGB1 in the induction of autoantibody formation in a TLR2-dependent manner, in the amplification of proinflammatory processes by HMGB1–DNA–autoantibody complexes via RAGE and TLR9 pathways, and in the local renal damage through binding to pathogenic anti-DNA autoantibodies in kidneys. Hence, HMGB1 and its receptors may represent a potential therapeutic target in SLE. Moreover, HMGB1 could be a valuable biomarker for SLE disease activity.