Development and histologic characterizations of an animal model of antigen-induced arthritis of the juvenile rabbit temporomandibular joint

Children with juvenile rheumatoid arthritis or juvenile chronic arthritis often exhibit temporomandibular joint (TMJ) involvement accompanied by pain, dysfunction, and growth abnormalities. Despite the severe functional and developmental consequences of this disease, its pathogenesis remains poorly understood, but important insights may be provided by a suitable animal model of this disease. The purpose of this study was to develop and histologically characterize a juvenile animal model of antigen-induced arthritis of the TMJ. Arthritis was induced with an intra-articular administration of ovalbumin in previously sensitized 10-week-old male New Zealand white rabbits. Sham-treated and untreated rabbits were used as controls. The TMJs were retrieved en bloc at 5, 10, 15, 35, and 55 days post-challenge for histology and matrix histochemistry. Antigen-treated joints demonstrated severe arthritis, including mononuclear cell infiltration, synovial lining and villous hyperplasia, and pannus formation, as early as 5 days after challenge; the arthritis was maintained up to 55 days post-challenge. A decrease in the area of the TMJ disc that stained positively for glycosaminoglycans was observed throughout the experimental period. Loss of collagen staining was primarily localized to sites at the junction of the synovium with bone and fibrocartilage. The histopathologic features of this model of antigen-induced arthritis of the juvenile rabbit TMJ are similar to those observed previously in adult animal models of experimental arthritis and in human rheumatoid arthritis. This animal model will be useful for understanding the pathogenesis of juvenile rheumatoid arthritis of the TMJ, and for exploring the mechanisms for aberrant craniofacial growth.     

Agency and communion: the relationship between therapy and culture

The detailed distribution and heterogeneity of various immunocompetent cells were characterized in the normal adrenal gland of the rat, with special emphasis on major histocompatibility complex (MHC) class II-expressing cells and macrophages. All adrenals contained at least two different populations of cells reactive with the dendritic cell or the macrophage antibodies. These cells were clearly distinguished from adrenal parenchymal cells by their morphology and location. The majority of dendritic cells were immunoreactive for the MHC class II (Ia) antigen (MRC OX6) and/or the dendritic cell antibodies (MRC OX62), and negative for the macrophage antibodies (ED1, ED2, and/or MRC OX42), whereas the main population of macrophages was immunonegative for the former antibodies and positive for the latter. The OX62-positive cells and the OX42-labeled cells occurred exclusively throughout the medulla. The cellular density of dendritic cells in the adrenal cortex was significantly higher than that of macrophages. Double-immunoperoxidase staining for ED1 and OX6 revealed that positively stained cells could be classified into the following categories: ED1+OX6+, ED1+OX6-, and ED1-OX6+. More then 40% of OX6+ cells were immunoreactive for ED1 in the zona glomerulosa, while approximately 15%, 20%, and 30% of OX6+ cells were positive for ED1 in the zona fasciculata, zona reticularis and medulla, respectively. ED1+ED2- cells were more frequently detected in the zona glomerulosa than in other adrenal zones. Only a few ED1-ED2+ cells were located in the zona glomerulosa, whereas a large number of them were found in the zona fasciculata. In the zona reticularis and medulla, ED1+ED2+, ED1+ED2-, and ED1-ED2+ cells were detected in the ratio 2:1:3. Our rsults suggest that dendritic cells and macrophages mature during their migration within the adrenal gland. These immunocompetent cells may contribute to a paracrine regulation of adrenal function under physiological conditions.     

Maturation of rat dendritic cells during intrahepatic translocation evaluated using monoclonal antibodies and electron microscopy

Specific populations of hepatic sinusoidal cells were stained with monoclonal antibodies that recognize monocytes/macrophages (ED1), tissue macrophages (Kupffer cells) (ED2), MHC class II (Ia) antigen (MRC OX6), and dendritic cells/gamma,delta T-cells (MRC OX62) and analyzed by light and electron microscopy. The majority of ED1(+) and/or ED2(+) cells were localized to the hepatic parenchyma, whereas OX6(+) and/or OX62(+) cells were more densely distributed within Glisson's sheath than in the hepatic parenchyma. Double-immunoperoxidase staining of normal liver for ED1, ED2, and OX6 identified dendritic cells (DC) of two different phenotypes, ED1(+)ED2(-)OX6(+) and ED1(-)ED2(-)OX6(+). DC can be classified into three different types based on ultrastructural characteristics. The first type (type I) is characterized by one or more long cytoplasmic processes and a well-developed lysosomal system. The second type (type II) has an inconspicuous lysosomal system, abundant hyaloplasm, and characteristic short cytoplasmic processes. The third type (type I-II) has cytologic features intermediate between those of type I and type II DC. At the electron-microscopic level, these three cell types are found in the sinusoidal lumen, whereas the majority of type II DC are located in the space of Disse and Glisson's sheath. Furthermore, some OX6-labeled elongated DC appeared to traverse the lumen of sinusoids through endothelial pores to enter the space of Disse. One hour after intravenous injection of latex particles (0.81 micrometer in diameter), numerous latex-laden dendritic cells (ED1(+)OX6(+), type I and type I-II) were detected in the lumen of hepatic sinusoids, but not in the space of Disse or Glisson's sheath. These findings suggest that normal rat liver contains resident dendritic cells which downregulate phagocytic activity and mature into potent accessory cells during migration from the portal vein toward the central vein. These DC then traverse the sinusoidal lumen to the hepatic lymph system via the space of Disse.     

Neural structures within the sheep temporomandibular joint

To better understand pathologic processes associated with arthritis of the temporomandibular joint (TMJ), detailed information on the innervation of TMJ tissues in normal as well as arthritic joints is needed. The aim of this study was to describe the normal innervation of the sheep TMJ in preparation for using this animal as a model for the study of the effects of arthritis on joint innervation. The macroscopic and microscopic appearance plus the distribution of neural structures within the TMJ were examined using fluorescence histochemistry (glyoxylic acid), immunohistochemistry (calcitonin gene-related peptide), silver, and gold chloride techniques. Joints from 10 mature merino sheep were studied. Calcitonin gene-related peptide-immunoreactive nerve fibers were found in the capsule and the synovial membrane, but not in the disc. Nerve bundles and single nerve fibers in the capsule, synovial membrane, and the peripheral 2 to 3 mm of the disc were stained by glyoxylic acid. Ruffini, paciniform-type, and Golgi organ nerve endings plus free nerve endings were located in the capsule, with the highest density of nerve endings occurring at the site of attachment of the disc to the capsule. The highest density of neural structures (using gold chloride) was in the posterior part of the joint. The highest density of autonomic fibers (using glyoxylic acid) was in the anterior capsule. The highest density of sensory fibers (using calcitonin gene-related peptide) was in the synovial and subsynovial tissues of the anterior capsule. These results confirm the existence of autonomic and sensory nerves in the capsule, synovial membrane, and peripheral disc in healthy adult sheep.     

Synovial inflammation in arthroscopically obtained biopsy specimens from the temporomandibular joint: a review of the literature and a proposed histologic grading system

Data indicate that the synovial lining of the temporomandibular joint (TMJ) in some respects differs from other joints. The normal variation in morphology of the synovial lining of the TMJ is quite great, whereas the variation in pattern of pathologic changes appears to be relatively small (ie, synovial inflammation is not of the severity as that in other joints). In the current review, a system for histologic grading of synovial inflammation is proposed. The system is based on semiquantitative evaluation of the following set of parameters: 1) synovial lining cell layers; 2) vascularity (number or size of vascular profiles); and 3) Inflammatory cell infiltrate (commonly lymphocytes).     

Pathogenesis and immunopathology of rheumatoid polyarthritis

Etiology of rheumatoid arthritis remains unresolved. Initiating mechanisms include a genetic background of susceptibility, defined by HLA-DR4 and -DR1 antigens and also hormonal factors and infectious events. The synovitis is the hallmark of the disease. Lymphocytes infiltrating the synovial membrane are mainly memory CD4+ T cells associated to dominant clonotypes. The proliferation and cytokine production by synovial tissue T lymphocytes are weak. B cells produce locally auto-antibodies: rheumatoid factors, anti-keratin antibodies.... Type A and B synoviocytes constitute an hyperplastic synovial lining. They synthesize and release large amount of proteolytic enzymes and proinflammatory cytokines, largely participating to inflammation and tissue destruction. Understanding the pathogenesis of rheumatoid arthritis should offer novel approaches to treat rheumatoid arthritis patients.     

Epidemiology of childhood disorders in a cross-cultural context

Platelet activating factor (PAF) is a potent mediator of allergic and inflammatory reactions in different pathological conditions. During recent years there has been increasing evidence that PAF can play an important role in the pathogenesis of arthritis. The PMN proteinases make an important contribution to the final tissue joint destruction in arthritis. In a rabbit model of acute crystal arthritis, we have compared the anti-inflammatory effect of two new molecules: BN 50727 with anti-PAF activity, and BN 50548 an inhibitor of PMN proteinases. These molecules were administered dissolved in DMSO at doses of 6 mg/kg three times daily i.p., beginning 24 h before the induction of arthritis. Compared with the untreated animals those receiving the drugs, presented a significant diminution in: (1) the synovial fluid volume; (2) the amount of cells infiltrating the joint cavity and the synovial membrane; and (3) the PGE2 concentration. Furthermore, in both groups of treated rabbits there was a significant decrease in synovial IL-6 concentration and in C-reactive protein serum levels and an important decline of histopathological score. The treatment with BN 50548 induced a significant reduction of TNF levels in the synovial fluid vs DMSO-treated and untreated rabbits. These results further strengthen that in an acute experimental arthritis model, molecules with capacity to antagonize the in vivo action of PAF have an anti-inflammatory effect reflecting an important role for this mediator in the pathogenesis of arthritis. We have also seen that an inhibitor of proteinases is capable of improving the joint inflammation apparently through a decrease in tumor necrosis factor (TNF) and interleukin-6 (IL-6) synovial levels. Furthermore, the proteinase inhibitor treatment preserves the loss of articular proteoglycan content, in an acute arthritis model. In conclusion, BN 50727 and BN 50548, two compounds with PAF antagonist and antiproteinase activity, respectively exert an anti-inflammatory effect in an experimental model of acute urate crystal arthritis, probably due to a decrease in TNF alpha and IL-6 synthesis.     

Responses of macrophage-associated antigen-expressing cells in the dental pulp of rat molars to experimental tooth replantation

Bacterial infection of the dental pulp is a major hindrance to successful pulp regeneration after tooth replantation. This study examined how macrophages and class II molecule-expressing cells of the pulp respond to tooth replantation, on the hypothesis that they contribute to the defence and repair of the traumatized pulp. Upper right first molars of 5-week-old male Wistar rats were replanted immediately after extraction; contralateral untreated teeth served as controls. Pulpal cells expressing macrophage-associated antigens were immunohistochemically demonstrated at 0 h (immediately after the replantation) to 84 days postoperatively using antirat monoclonal antibodies OX6 (anti-class II molecules), ED1 (pan-macrophage antibody, reactive also with dendritic cells) and ED2 (anti-resident macrophages). Between 3 and 7 days postoperatively, ED1+ and OX6+ cells, but not ED2+ cells, were concentrated in areas of degeneration formed in the coronal pulp, and frequently showed a marked accumulation along the pulp-dentine border of the cuspal area. Confocal laser scanning microscopy revealed that some of the OX6+ cells with a dendritic profile extended several cytoplasmic processes into the dentinal tubules communicating with the enamel-free area at the tip of the cusp. From 14-84 days, approx. two-thirds of specimens exhibited pulp-tissue regeneration with increasing formation of reparative dentine. Following the formation of sound reparative dentine, cells positive to each antibody were distributed more centrally in the pulp than in the controls, and thus did not show any accumulation along the pulp-dentine border. However, in the other specimens where a bone-like hard tissue had formed in the pulp chamber, many ED1+ and OX6+ cells were still concentrated in the remaining pulp tissue and showed a marked accumulation along the pulp dentine border. Few ED2+ cells were observed in these specimens. These findings suggest that, following tooth replantation, exudative macrophages are actively engaged in eliminating dentinal tubule-derived infectious stimuli and that class II molecule-expressing cells, most probably containing dendritic cells, are positioned strategically at the outermost portion of the injured pulp to monitor incoming antigens. The intensity of the pulpal defence reaction may be dependent on the status of hard-tissue formation, which influences the amount of incoming antigens.     

Increased expression of human type IIa secretory phospholipase A2 antigen in arthritic synovium

OBJECTIVE: To determine the localisation and level of expression of human type IIa secretory phospholipase A2 (sPLA2) in the synovium of rheumatoid arthritis (RA), osteoarthritis (OA), and non-arthritic (NA) patients and to examine the relation between sPLA2 and histological features of inflammation. METHODS: Immunoperoxidase staining using the anti-sPLA2 monoclonal antibody 9C1 was performed on frozen sections of knee synovium of 10 RA, 10 OA, and 10 NA patients. sPLA2 positive cells were scored on a scale of 0-3 in 10 fields of a representative tissue section from each case. Double labelling immunofluorescence confocal microscopy with antibodies to CD14 or CD45 and 9C1 was used to determine cell type specificity. Inflammation was assessed by semiquantitative scoring of lining layer thickness and mononuclear cell infiltrates (MC) and a cumulative inflammation score, generated by summing the two parameters. Scores in each group were compared using non-parametric statistical analysis. RESULTS: sPLA2 was localised to endothelium (EC), vascular smooth muscle (VSM), and mast cells (M) in all tissue sections. In RA and OA sections, staining was seen in both macrophage-like and fibroblast-like cells in the synovial lining layer (LL) and subsynovial lining layer (SLL). Perineural cells stained positively. Subintimal lymphoid aggregates (LA) were negative in all sections. The RA group showed significantly greater staining in extravascular synovial tissue (median 3.6, range 1.5-6.0) than the OA (median 1.95, range 0-5.3) or NA (median 0, range 0-5.9) groups (p < 0.05). LL staining was significantly higher in RA than both OA and NA sections (p < 0.05). The OA group showed a trend to higher staining scores than the NA group that did not reach significance. There was a significant correlation between the sPLA2 staining score and inflammation score within the RA patient group (p < 0.05). CONCLUSIONS: The synovium is a site of increased expression of sPLA2 antigen in both RA and OA relative to NA. Its presence in both fibroblast and macrophage-like cells in the LL and SLL of synovial tissue in RA and OA, but not NA, indicates that the enzyme is specifically induced in these regions in both conditions with expression in the LL being particularly characteristic of RA. The widespread expression of sPLA2 in synovium suggests it is likely to play a significant part in synovial pathology.     

Inflammatory infiltrate and interleukin-8 expression in the synovium of psoriatic arthritis--an immunohistochemical and mRNA analysis

Psoriatic arthritis is an inflammatory arthropathy that ultimately can lead to joint destruction. In this study, we investigated the immunophenotypes of the inflammatory cells and the expression of interleukin-8 (IL-8), which is the hallmark chemoattractant cytokine of psoriasis in synovial membranes from patients exhibiting active psoriatic synovitis (n = 9). The tissue samples were examined by immunohistochemistry, Western blot analysis and in situ hybridisation. The inflammatory infiltrate consisted predominantly of CD3+ T lymphocytes, with a higher proportion of CD4+ than CD8+ T lymphocytes in six cases. CD3+ T lymphocytes were focally distributed near small blood vessels and the enlarged synovial intima. CD1+ interdigitating reticulum cells were not detected. CD22+ B lymphocytes and plasma cells were found in small aggregates without KiM4+ follicular dendritic cells. KiM8+ macrophages were located in the synovial intima and were distributed in a diffuse pattern near the synovial lining cells. CD15+ neutrophil granulocytes were detected in four cases. They were preferentially located in the vicinity of blood vessels and the synovial intima. IL-8 was found at a high level in the synovial lining cells and to a lesser extent in cells located in the perivascular areas. Immunofluorescence double staining showed IL-8 to be expressed in KiM8+ multinucleated giant cells, KiM8+ macrophages and CD3+ T lymphocytes. IL-8 receptor A was demonstrated in the synovial lining and in macrophages and lymphocytes. IL-8 was detected by immunoblot analysis of the synovial tissue at 8.4 kD. Employing in situ hybridisation, IL-8 mRNA was strongly and preferentially expressed in the synovial intima, as well as in macrophages and lymphocytes. The immunophenotype of the psoriatic arthritis inflammatory cells shows great similarity to the inflammatory infiltrate found in the synovial tissue of patients with rheumatoid arthritis. The preferential expression of IL-8 and IL-8 mRNA in the enlarged synovial intima and in lymphocytes and macrophages suggests that IL-8 exerts its action through activated mononuclear cells and T lymphocytes. It seems to play a role in regulating leucocyte traffic into the enlarged synovial intima and may contribute to the aggressive synovitis of patients with psoriatic arthritis.     

Expression of integrin family of cell adhesion receptors in rheumatoid synovium. Alpha 6 integrin subunit in normal and hyperplastic synovial lining cell layer

Integrins are heterodimeric cell adhesion receptors. The beta 1 integrin subunit can be in a complex with multiple a subunits and form receptors for collagen, laminin, fibronectin, and vitronectin. We have characterized the distribution of eight integrin subunits in rheumatoid synovium, with special interest in the lining cell layer. The beta 1 integrin subunit was found in abundance in synovial stroma and in lining cells. The only alpha subunit seen constantly in lining cells was alpha 6. In complex with alpha beta subunit, alpha 6 forms a laminin receptor usually seen in epithelial or endothelial cells or in macrophages. The fact that laminin was found in the extracellular matrix around synovial cells suggests the importance of alpha 6 integrin in the adhesion of synovial lining cells. Furthermore, alpha 6 expression was noticeably weaker in strongly proliferative lining cell layers, indicating that the inflammatory process may regulate integrin expression. A potential connection between altered expression of cell adhesion receptors and the pathological behavior of rheumatoid lining cells is suggested.     

Molecular lysis of synovial lining cells by in vivo herpes simplex virus-thymidine kinase gene transfer

Herpes simples virus thymidine kinase (HSV-TK) expression plasmid DNA was injected into the joint space of rabbits with antigen-induced arthritis (AIA). Purified plasmid DNA was able to mediate transfection of synovial lining cells and transient overexpression of HSV-TK in the context of active synovial inflammation. The pharmacodynamic distribution of intraarticular expression plasmid DNA was confined to the joint space. Arthritic rabbits treated with intraarticular expression plasmid DNA followed by intravenous ganciclovir (GCV, 5 mg/kg) twice daily for 3 days showed histologic evidence of synovial lining layer cytolysis when articular tissues were examined 21 days posttreatment. There was also a reduction in joint swelling in the TK-treated knees. No untoward clinical effects were observed in the rabbits and no evidence of cytolytic damage specific to the TK-GCV gene therapy was observed either in the articular cartilage or bone. The application of TK-GCV intraarticular gene therapy using purified expression plasmid DNA for the induction of synovial cytolysis may be applicable to the treatment of human inflammatory arthritis.     

Conformational change of helix G in the bacteriorhodopsin photocycle: investigation with heavy atom labeling and x-ray diffraction

Collagen-induced arthritis in DBA/1 mice is a widely used experimental model of rheumatoid arthritis. The induction phase of the disease is thought to be dependent upon MHC-restricted T and B cell-mediated immune responses to type II collagen, but an influence of additional non-MHC-restricted mechanisms has also been proposed. In this study, we report that type II collagen immunization of DBA/1 mice lacking mature T and B lymphocytes resulted in the development of arthritic lesions, which were characterized by synovial hyperplasia with occasional inflammation as well as cartilage and bone destruction. The specificity of disease induction to type II collagen was confirmed, because arthritis could not be induced when control preparations of OVA or adjuvant alone were administered. A delay in clinical disease onset and a reduction in severity between lymphocyte-positive and -negative DBA/1 mice confirmed that lymphocytes play an important role in disease; however, similar pathologic features and normal incidence suggest that lymphocyte-independent mechanisms of disease induction also operate in the standard collagen-induced arthritis model. We conclude that adaptive immune responses are not the only arthritogenic mechanism and hypothesize that the nonantigenic properties of type II collagen can also lead to arthritis.     

Pannus development in adjuvant arthritis of the rat - an autoradiographic study (author's transl)

After the injection of complete adjuvant, rats develop arthritis with increased proliferation of synovial tissue cells at the attachment of the synovial membrane to cartilage and bone. From this synovial tissue a connective tissue pannus develops which covers the cartilage surface as a monocellular or multicellular layer of proliferating cells. Areas with cartilage destruction are usually characterized by a great number of proliferating connective tissue cells. Cartilage destruction is not restricted to the superficial pannus, since granulation tissue of the subchondral bone may also participate in cartilage resorption. The quantitative determination of 3H-thymidine labelled chondrocytes excludes participation of these cells in pannus formation.     

Expression of N-acetyl-D-galactosamine associated epitope in synovium: a potential marker of glycoprotein production

OBJECTIVE: To investigate synovial glycoprotein production in situ, a novel monoclonal antibody (Mab), A13D8, was used to evaluate the expression of an epitope containing N-acetyl-D-galactosamine (GalNAc) in normal and pathological synovium. METHODS: Immunohistological and cytochemical analysis of synovial tissue samples was undertaken with single and double staining techniques using the A13D8 Mab, anti-CD68, vascular cell adhesion molecule-1 (VCAM-1), the hyaluronan associated enzyme uridine diphosphoglucose dehydrogenase (UDPGD), and the anti-Golgi Mab SSN/HR-1992. The specificity of the A13D8 Mab was established through blocking studies using carbohydrate residues, including GalNAc and N-acetylglucosamine (GlcNAc). RESULTS: A13D8 is expressed intensely in the cytoplasm of normal type B lining cells, which coexpress VCAM-1 and UDPGD, and is not expressed by CD68+ type A lining cells. In the lining layer of RA synovium, there is a negative correlation between A13D8 expression and the level of lymphocytic infiltration in the sublining areas (r = -0.43, p < 0.001). The endothelium of a subset of venules, typically in lymphocyte-rich aggregates, also stains intensely for A13D8. Pretreatment of the Mab with GalNAc completely eliminates the tissue staining, as well as the 110 kDa band seen on immunoblot, whereas pretreatment of A13D8 with GlcNAc and lactose has no effect. Double staining of HEp-2 cells with A13D8 and the anti-Golgi Mab SSN/HR-1992 reveals co-localization of the A13D8 epitope to the Golgi apparatus. CONCLUSION: Type B synovial lining cells and selected synovial endothelium express GalNAc containing epitope identified by Mab A13D8. Marked reduction in the expression of this epitope in the lining layer of inflamed RA synovium suggests that the synovial production of GalNAc containing glycoproteins, such as mucins, may be altered in this disorder.     

Immunohistochemical distribution of lymph capillaries and blood capillaries in the synovial membrane in cases of internal derangement of the temporomandibular joint

The distribution of lymph capillaries and blood capillaries in the synovial membrane was examined immunohistologically with anti-human collagen IV antibody and anti-human von Willebrand factor in 26 human temporomandibular joint (TMJ) samples comprising discs with adjoining synovial membrane from 10 control TMJs and from 16 TMJs with internal derangement. Three different distribution types were observed in the synovial membrane. In the control samples, the occurrence of blood capillaries and lymph capillaries was rare. In mildly hyperplastic synovitis, lymph capillaries were observed just beneath the surface of the synovial membrane, whereas blood capillaries occurred in a little deeper layer of the synovial membrane. In a severely hyperplastic synovitis, both lymph and blood capillaries were observed frequently. The present results suggest that the different distribution patterns of lymph capillaries and blood capillaries reflect the degree of synovitis but can not be attributed to specific clinical symptoms.     

Effects of pulse methylprednisolone on cell adhesion molecules in the synovial membrane in rheumatoid arthritis. Reduced E-selectin and intercellular adhesion molecule 1 expression

OBJECTIVE: To investigate the effects of a 1,000-mg intravenous pulse of methylprednisolone succinate (MP) on cell adhesion molecule expression on the synovial vascular endothelium in patients with rheumatoid arthritis (RA). METHODS: Sequential arthroscopic biopsy samples were taken before and 24 hours after MP administration (10 patients) and at the time of RA flare (2 patients) and after retreatment with MP (1 patient). Immunoperoxidase staining for E-selectin (CD62E), P-selectin (CD62P), intercellular adhesion molecule 1 (ICAM-1; CD54) and platelet-endothelial cell adhesion molecule (PECAM; CD31) was performed, and the staining was quantified by color video image analysis. RESULTS: MP caused a rapid (within 24 hours) and substantial decrease in the expression of E-selectin on the synovial vascular endothelium, with a smaller reduction in ICAM-1 expression on synovial vascular endothelium and the synovial lining. There were no similar effects on synovial membrane P-selectin or PECAM expression. CONCLUSION: A potential mechanism by which MP impairs neutrophil trafficking into inflamed RA joints might be by reducing E-selectin, and possibly, ICAM-1, expression in the synovial membrane.