High mobility group box chromosomal protein 1 (HMGB-1) is a ubiquitous chromatin component expressed in nucleated mammalian cells. It has recently and unexpectedly been demonstrated that stimulated live mononuclear phagocytes secrete HMGB-1, which then acts as a potent factor that causes inflammation and protease activation. Macrophages play pivotal roles in the pathogenesis of arthritis. The aim of this study was to determine whether synovial macrophage expression of HMGB-1 is altered in human and experimental synovitis.
Intraarticular tissue specimens were obtained from healthy Lewis rats, Lewis rats with Mycobacterium tuberculosis–induced adjuvant arthritis, and from patients with rheumatoid arthritis (RA). Specimens were immunohistochemically stained for cellular HMGB-1. Extracellular HMGB-1 levels were assessed in synovial fluid samples from RA patients by Western blotting.
Immunostaining of specimens from normal rats showed that HMGB-1 was primarily confined to the nucleus of synoviocytes and chondrocytes, with occasional cytoplasmic staining and no extracellular matrix deposition. In contrast, inflammatory synovial tissue from rats with experimental arthritis as well as from humans with RA showed a distinctly different HMGB-1 staining pattern. Nuclear HMGB-1 expression was accompanied by a cytoplasmic staining in many mononuclear cells, with a macrophage-like appearance and an extracellular matrix deposition. Analysis of synovial fluid samples from RA patients further confirmed the extracellular presence of HMGB-1; 14 of 15 samples had HMGB-1 concentrations of 1.8–10.4 μg/ml.
The proinflammatory mediator HMGB-1 was abundantly expressed as a nuclear, cytoplasmic, and extracellular component in synovial tissues from RA patients and from rats with experimental arthritis. These findings suggest a pathogenetic role for HMGB-1 in synovitis and indicate a new potential therapeutic target molecule.
High mobility group box chromosomal protein 1 (HMGB-1; previously called high mobility group 1 [HMG-1] or amphoterin) is an intranuclear factor that facilitates protein interactions with chromatin (1). Hmgb1 knockout mice die shortly after birth because of hypoglycemia secondary to insufficient glucocorticoid receptor expression, which is under HMGB-1–mediated transcriptional control (2). HMGB-1 is ubiquitously present in the nucleus of almost all mammalian cells and is highly conserved between species (3). Beyond this intranuclear role, it has recently been discovered that HMGB-1 is secreted by certain cells, including activated monocytes and macrophages, and plays important roles in inflammation and tumor metastasis (4, 5). The molecule is a late mediator of endotoxin lethality in mice and can be successfully targeted by neutralizing antibodies (4). Moreover, injection of HMGB-1 was shown to induce toxic shock. Monocyte/macrophages activated by endotoxin, tumor necrosis factor (TNF), or interleukin-1 (IL-1) secrete HMGB-1 in substantial amounts (4). Furthermore, we have recently reported that extracellular HMGB-1 is a potent proinflammatory molecule that induces TNF and IL-1 synthesis and secretion in monocyte/macrophages. HMGB-1 causes a delayed and biphasic release of TNF compared with that seen after endotoxin activation (6).
It has recently been demonstrated that necrotic cells passively release their nuclear HMGB-1 extracellularly, thereby initiating a proinflammatory response (7, 8). Necrotic cells from Hmgb1−/− mice have a greatly reduced ability to promote inflammation, proving the need for HMGB-1 as a mediator of necrosis-induced inflammation. In contrast, in apoptotic cells that do not trigger inflammation, HMGB-1 is tightly bound to the chromatin due to generalized histone underacetylation, which prevents extracellular release. Thus, inflammation caused by necrosis or by monocyte/macrophage activation has a common molecular denominator in HMGB-1.
HMGB-1 binds with high affinity to the receptor for advanced glycation end products (RAGE), which is expressed on many cells, including phagocytic mononuclear cells (9, 10). Interaction between HMGB-1 and RAGE activates several intracellular signal transduction pathways, including nuclear factor κB (11), the transcription factor that is classically linked to inflammatory processes.
TNF and IL-1 play central roles in the pathogenesis of arthritis. Since HMGB-1 is a potent stimulus for the synthesis of both these factors, we decided to study whether HMGB-1 is expressed and released in the joints of patients with rheumatoid arthritis (RA) or in animals with experimental arthritis. Putative cellular sources for the intraarticular release of HMGB-1 in synovitis could be activated monocyte/macrophages and cells rendered necrotic by hypoxia or activated complement.
MATERIALS AND METHODS
Synovial tissue from animals with experimental arthritis.
Female Lewis rats weighing 175–200 gm were purchased from M&B (Ry, Denmark) and maintained at the animal unit at Karolinska Institutet. The light/dark cycle was 12 hours, and the rats were fed standard rodent chow and water ad libitum. The health status of the animals was monitored according to the guidelines of the Swedish Veterinary Board, and animals were reported to be free of the screened pathogens. All experimental procedures were approved by the Ethical Committee Stockholm North, Sweden.
On day 0, rats were immunized subcutaneously at the base of the tail with 0.5 mg of heat-killed Mycobacterium tuberculosis (strains C, DT, and DN; Central Veterinary Laboratory, Weybridge, UK) dispersed in 50 μl of paraffin oil. With this method, arthritis is expected to develop in 100% of the animals, with a clinical onset around day 11 after immunization. Arthritis was evaluated by measuring paw swelling with a plethysmometer. On days 0 and 30 after immunization, rats were perfused in vivo with formaldehyde solution, then paws were dissected and decalcified as previously described (12).
Synovial tissue and synovial fluid from patients with RA and osteoarthritis (OA).
Synovial biopsies were obtained from 3 RA patients and from 1 OA patient requiring joint replacement surgery. The RA patients fulfilled the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 criteria (13) and had active disease. There were 2 men ages 55 years and 80 years, with a disease duration of 5 months and 8 weeks, respectively. One was treated with a nonsteroidal antiinflammatory drug (NSAID), and the other was receiving an NSAID plus sulfasalazine. The other RA patient was a 40-year-old woman with a disease duration of 8 years; she was being treated with methotrexate.
The OA patient was a 60-year-old man with a disease duration of 8 years; he was receiving treatment with NSAIDs. The presence of OA was assessed radiographically according to the Kellgren/Lawrence method (14).
The tissue samples were immediately frozen in a dry-ice bath and stored at −80°C. After cryosectioning, the tissue specimens were fixed in a 2% formaldehyde solution.
Synovial fluid samples were collected from an additional 15 RA patients who fulfilled the ACR 1987 criteria. All 15 of these RA patients had active disease (Table 1).
Table 1. HMGB-1 and TNF levels in synovial fluid from RA patients*
Sex/age of the study patients
Duration of RA, years
On the day of blood sampling, patients were taking 7.5–15 mg of methotrexate (MTX) once a week, ≤7.5 mg of prednisone (pred.) per day, or 2 gm of sulfasalazine (SSZ) per day. HMGB-1 = high mobility group box chromosome protein 1; TNF = tumor necrosis factor; RA = rheumatoid arthritis.
Synovial tissue sections (8 μm thick) were cut with a cryostat (Leica, Cambridge, UK) and mounted on ProbeOn microscope slides (Fisher Scientific, Pittsburgh, PA). To detect the presence of cellular and extracellular HMGB-1, sections were stained with a peptide affinity-purified polyclonal rabbit anti–HMGB-1 antibody (catalog no. 556528; PharMingen, San Diego, CA) according to methods we have previously described (15, 16). Human tissue samples were also stained with a mixture of 2 mouse monoclonal antibodies (mAb) directed against TNF (mAb 1 and mAb 11; PharMingen). Saponin was not included in the staining protocol for HMGB-1 in rat tissue.
Sections were then counterstained with Mayer's hematoxylin and mounted with buffered glycerol. Slides were evaluated using a Polyvar II microscope (Reichert-Jung, Vienna, Austria) connected to a charge-coupled device color camera (DXC-750P; Sony, Tokyo, Japan).
The presence of HMGB-1 in synovial fluid was detected by fractionation of synovial fluid by denaturing sodium dodecyl sulfate–polyacrylamide gel electrophoresis, followed by Western blotting. HMGB-1 levels were determined by comparison against a standard curve of purified recombinant HMGB-1 diluted in saline, using a gel image analysis system (4). The TNF content was analyzed by a commercial TNF enzyme-linked immunosorbent assay kit obtained from R&D Systems (Minneapolis, MN).
Expression of HMGB-1 in normal and arthritic rat paws.
Immunohistochemical staining was performed to study the expression of HMGB-1 in normal synovial tissue from healthy rats compared with synovial tissue from rats with adjuvant-induced arthritis. We identified HMGB-1 expression in both normal and inflammatory synovial specimens. The cellular localization of HMGB-1 differed, however. Joint tissues from healthy rats revealed a nuclear location of HMGB-1 in most cells in the synovial membrane and in many chondrocytes in the superficial area of the articular cartilage (Figures 1A and B). Only occasional cells expressed cytoplasmic HMGB-1, and no deposition of HMGB-1 in the extracellular matrix was seen.
In contrast, HMGB-1 expression in inflamed synovial tissues was also detectable cytoplasmically in mononuclear cells with a macrophage-like appearance (Figures 1C and D). Similar macrophage-like cells in the synovial fluid from arthritic rat joints also expressed HMGB-1 cytoplasmically, while adjacent polymorphonuclear cells were negative (Figure 1E).
Expression of HMGB-1 in synovial tissue from RA.
Synovial biopsy specimens from patients with active RA were sectioned and immunostained to study the cellular expression pattern of HMGB-1. Macrophage-like cells in RA biopsy specimens expressed HMGB-1 both cytoplasmically and intranuclearly in a manner similar to that seen in the experimental animal model (Figure 2A). In the OA patient evaluated, HMGB-1 expression in the synovial tissue biopsy specimen was mainly nuclear, and only occasional cells expressed cytoplasmic HMGB-1 (Figure 2C).
Extracellular release of HMGB-1 into synovial fluid.
Synovial fluid collected from an additional 15 RA patients was analyzed for the presence of extracellular HMGB-1 and TNF (Table 1). HMGB-1 could be demonstrated in the synovial fluid of 14 of the 15 patients, with levels ranging from 1.8 μg/ml to 10.4 μg/ml. TNF was detected in 12 of the 15 patients, with levels ranging from 8 pg/ml to 1,608 pg/ml. No correlation between HMGB-1 levels and TNF levels was found.
This is the first report demonstrating extracellular and cytoplasmic expression of HMGB-1 in the context of a chronic inflammatory disease. Recent evidence indicates that HMGB-1 is an important proinflammatory molecule, mediating septic shock (4), acute lung injury (17), smooth muscle chemotaxis (18), and potent TNF/IL-1 release from activated macrophages (6). In both septic shock and acute lung injury, treatment with anti–HMGB-1 ameliorated the inflammation. In the case of lung injury, both reduced neutrophil migration and reduced edema formation were seen. HMGB-1 is not only secreted by macrophages in response to proinflammatory stimuli, but the molecule itself provokes a response that prolongs and sustains inflammation. Macrophages play a pivotal role in the pathogenesis of arthritis in humans and in experimental models. We therefore sought to determine whether abnormal HMGB-1 expression could be demonstrated in chronic arthritis, which would suggest a disease-promoting role for this molecule.
Since HMGB-1 is a ubiquitously expressed nuclear factor in all mammalian cells, we had expected to demonstrate the protein in all cell nuclei in the normal tissue specimens. However, not all cells stained positively for HMGB-1. This might be a result of insufficient sensitivity of our method to visualize nuclear HMGB-1 expression at low levels. A strikingly different HMGB-1 staining pattern, both quantitatively and spatially, was demonstrated in inflammatory synovial tissue from rats with adjuvant-induced arthritis as well as from humans with RA. The nuclear HMGB-1 staining was accompanied by cytoplasmic staining in many rounded mononuclear cells with a macrophage-like appearance. Furthermore, brownish staining of the extracellular matrix of inflamed synovium indicated the release of cytoplasmic HMGB-1, either from activated macrophages or from necrotic cells. Analysis of synovial fluid levels in RA patients further confirmed the extracellular presence of HMGB-1 in the majority of the samples examined.
These findings thus demonstrate that intracellular HMGB-1 can be released during inflammation, which is a prerequisite for HMGB-1 to interact with its high-affinity cellular receptor RAGE. This receptor is expressed on a wide variety of cell types including endothelial cells, mononuclear phagocytes, neurons, and smooth muscle cells.
Ligand–RAGE interactions have been implicated in several pathologic processes, such as diabetes mellitus, amyloidosis, and atherosclerosis (for review, see ref. 10). Our results suggest that the interaction between HMGB-1 and RAGE may also be a pathogenetic factor in the development of arthritis. The unexpectedly increased levels of synovial fluid HMGB-1 seen in most RA samples correspond to levels we have previously found to lead to strong TNF and IL-1 formation in cultured macrophages and monocytes (6). However, in the present study, we found no correlation between the synovial fluid levels of HMGB-1 and TNF as determined by enzyme-linked immunosorbent assay. Further studies will be required to clarify this discrepancy.
Extracellular HMGB-1 is also known to bind to several components of the plasminogen activation system and to enhance the activity of tissue plasminogen activator (19) and matrix metalloproteinases 2 and 9 (5). It is thus plausible that HMGB-1 plays a role in the inflammatory process as well as the destructive process in the pathophysiology of arthritis.
Since the administration of anti–HMGB-1 antibodies prevented lipopolysaccharide-induced lethality in mice (4), HMGB-1 neutralization might be beneficial in the treatment of inflammatory diseases. We thus speculate that HMGB-1 provides a potential target for therapeutic intervention in arthritis. Studies are in progress in our laboratory to explore this potential.
We are grateful to Ann-Britt Wikström, Lotta Aveberger, and Tomas Ernemar for excellent technical assistance and to Associate Professor R. A. Harris for critically reading the manuscript.