1430 AKIIIRITIS & RIIEUMATISM Vol. 39, No. 8, August 1996, pp 1430-1434 0 1996, American College of Rheumatology CONCISE COMMUNICATIONS Modulation of human chondrocyte integrins by inflammatory synovial fluid Cell-matrix interactions, mediated by integrins, are increasingly important in the regulation of chondrocytc function and development (1-3). These interactions may be modified by synovial fluid (SF)-derived cytokines and growth factors in osteoarthritis (OA) and rheumatoid arthritis (RA). SF cytokines regulate the production of matrix macromolecules and protcases, but less is known of their effects on integrin expression by articular chondrocytes. We have thcrefore examined the expression of integrins in RA cartilage in situ and investigated the effect of inflammatory SF on integrin expression by human articular chondrocytes in vitro. Full-depth cartilage, dissected from medial tibial plateaux of 3 cartilage specimcns obtained from patients with RA at joint replacement surgery, was snap-frozen, cut, and stained with the following antibodics: TS2/16 (anti-pllCD29; Dr. F. Sanchez-Madrid, Universidad Authomale, Madrid, Spain); MHM 23 (anti-PUCD18, Dako, Santa Barbara, CA); YS/61 (antiP3/CD61; Dako); 3E1 (anti-p4; Chemicon International, Danvers, MA); TS2/7 (anti-al/CD49a; T Cell Sciences, Cambridge, MA); Gi9 and Gi14 (anti-a2/CD49b; Dr. S. Santoso, Justus Liebig University, Giessen, Germany); P1B5 and J143 (anti-a3/CD4Yc; Gibco, Grand Island, NY,and Dr. S. Dedhar, University of Toronto, Toronto, Ontario, Canada, rcspcctively); HP2/1 (anti-a4/CD49d; Dr. F. Sanchez-Madrid); SAM-1, PlD6, BIIG2 ( a n t i d ; Serotcc, Kidlington, Oxford, UK, Gibco, and Dr. C. Damsky, Univcrsity of California, San Francisco, CA, respectively); and GOH3 (anti-a6/CD4Yf; Serotec), as described (4). Antibody binding was detected by an avidin-biotin-conjugated peroxidasc technique. Sections of normal human tonsil served as positive controls. Primary antibodies were omitted in negative controls. RA tissue sections showed severe loss of cartilage as well as secondary OA. Chondrocytes in all zones stained with anti-pl antibodies but showed no staining for p2, /33, or p4. All chondrocytcs stained strongly for a5, with no focal differenccs. Expression of a1 and a 3 was variable. Many cells showed weak or no immunoreactivity for these subunits, but in areas where chondrocyte clusters were present, there was strong expression of a1 and a3. Chondrocytcs showed no immunoreactivity for 122, a4,or a 6 integrins. Articular chondrocytes were isolated from tibial plateau, femoral condyle, or humcral head articular cartilage obtained at autopsy from 24 individuals who had died of causes unrelated to joint diseasc, by digestion with type IA collagcnase (1.5-3 mg/ml; Sigma, St. Louis, MO) for 6-16 hours, depending on the tissue volume. Isolated cells were cultured for 48 hours, either in monolayers or encapsulated in alginate beads ( 5 ) , in Ham's F-12 supplemented with antibiotics, 50 &ml of ascorbic acid, and 2 mM L-glutamine (all from Sigma) under serum-free conditions or with the addition of 20% normal human serum or 20% SF. The SF sample was obtained from a patient with active RA. It was cell-free and had been treated with hyaluronidase (30 units/ml for 30 minutes at 37°C). Results obtained with other inflammatory SFs were similar to those dcscribcd below. Monolayer-cultured cells were released with trypsin- EDTA (6), while cells in alginate bead cultures were released non-enzymatically (5). Stimulated chondrocytes were stained with the following anti-integrin antibodies: TS2/16, TS2/7, PlB5, PlD6, and 13C2 (anti-aV/CDSl; Dr. M. Horton, University College London Medical School, London, UK), as previously described (6). Antibodies against a-fctoprotcin (Prof. V. van Heyningen, Medical Research Council, Edinburgh, UK) and keratan sulfate (MZlS; Dr. M. Bayliss, Kennedy Institute, London, UK) scrved as negative and positive controls, rcspectivcly. Sccondary staining was with a goat anti-mouse phycocrythrin conjugatc (Sigma). Cell viability was assessed with propidium iodide (5 mg/ml). Staining was cxamined by FACScan and Lysis I1 software (Bccton Dickinson, Mountain View, CA), on 20,000 events with stable acquisition parameters (6). Staining intensity was compared using the Wilcoxon signed rank test. Culture of cells in monolayer with serum or SF led to an increase in expression of all intcgrins examined when compared with cells cultured under serum-free conditions (Figure 1). Expression of integrins a5, aV, and pl was more intense than the expression of a1 and a3 under all culture conditions ( P 5 0.05). Mean fluorescence intensity values (2SEM) in 7 experiments were: negative control 3.6 % 1.3, MZ15 81.1 ? 12, 01 8.6 ? 2.5, a 3 8.1 ? 1, a5 15.1 % 4.3, aV 17.6 2 3.3, and /3l 23.9 2 5.8. In monolayer cultures, SF induced a significant incrcasc in thc cxprcssion of as, a l , and aV cornpared with integrin cxprcssion under serum-free conditions, whereas serum induced only a5 (Figure 1). Chondrocytes cultured in alginate beads also showed increased a1 and a5 when cultured with SF but no significant change with serum. Intcgrin staining intensity was grcatcr in monolayer cultures than in alginate cultures (mean 2 SEM % difference, a1 4 ? 9%, a 3 65 ? 3096, a5 14 ? 7%, a V 28 ? 9%, and pl 31 % 8%; n = 3; P < 0.001). Cell viability was not affected by culture conditions. Increased expression of 01 and a 3 has also been observed in chondrocyte clusters in situ in experimcntal primate OA (7). Apparcnt incrcascs may, howcvcr, rcsult from a higher concentration of cells or better penetration of antibodies into tissues, rather than a true increase in expression. Possible explanations of a genuine increase have been discussed by Locscr el a1 (7). Cartilage, at an early stage of RA, is rarely available for examination, but such tissue may represent the rheumatoid proccss more accurately than tissue obtained at joint replacement surgery. Studies with isolated chondrocytes indicate that the culture methods and the presence of serum or SF can influcncc the expression of certain integrin subunits. Differences in the effects of serum and SF on the expression of a l , aV, and a5 in monolayer cultures may reflect the presence of one or more cytokines in SF or may be a consequence of differences in the quantity or activation state of growth factors such as transforming growth factor /3 (3). Indeed, the activation state of the integrin receptors themselves may be altered by factors in serum or SF, and this could influence the binding of the anti-integrin antibodies used in this study. Presumably, enhanced expression induced by serum or SF occurs in parallel with changes in matrix synthesis. The cellular distribution of 05/31 (fibronectin receptor), strongly expressed by articular chondrocytes, may be altered by expo- 1431 CONCISE COMMUNICATIONS 400 0 Monolayer cultures: 20% NHS Monolayer cultures: 20% SF * p Alginatc culaucs: 20%NHS m- 200 * i** 0 I a1 a3 a5 aV B1 -100 Figure 1. Effect of normal human serum (NHS; 20%) and inflammatory synovial fluid (SF 20%) on selected intcgrin subunit expression by human articular chondrocytes.The data show a mcan (tSEM) percentage increase in mean fluoresence intensity compared with the expression under serum-free conditions. * = P < 0.02; ** =- P < 0.04; *** P = 0.0078. A minimum of 5 cxperiments were carried out in each categoq, except where p1 cxprcssion was examined in monolayer cultues, where n = 4. ~ sure to ligand (8). Thus, intact fibronectin in serum or SF, and proteolytic fragments of fibronectin in SIT, could mediate changes in a5 expression. The significance of the altered expression of al, as induced by SF in all cultures and seen with greater intcnsity in chondrocyte clustcrs in situ, is unknown. This could reflect increased alp1, a chondrocytc receptor for collagens (7), perhaps as a consequence of altcrcd collagen synthesis since interleukin-1, which is present in abundance in RA SF,induces chondrocytes to synthesize collagen types I and 111. Cell attachment alone can induce the expression of a 5 p l (9), but SF also induced expression of a1 and a5, but not aV, integrin in the alginate bead cultures. The changes were, however, less in alginate cultures. This suggests that cell shape is an important determinant of expression, especially of aV. This idea is supported by the observation that aV is more intensely expressed in superficial chondrocytcs (lo), which are flattened, compared with deep-zone cells, which are round. Induction of aV by SF in monolayer cultures probably reflects increased aVP5 expression, since we did not detect 83 in situ (thus excluding aVp3 heterodimer), and aVp1 was not found by Woods et al (10). Potential ligands for aV integrins include fibronectin, vitronectin, and ostcopontin. Accumulation of fibronectin is well documented in diseased cartilage, but little is known about expression of vitronectin and osteopontin. Induction of d p l , in particular, by SF probably leads to an increase in the number of binding sites for fibronectin, and this may be accompanied by accumulation of fibronectin in dis- eased cartilagc. It is known that skin fibroblasts accumulate more fibronectin when binding sites are increased (9). Conclusive evidence that modulation of chondroeyte integrin expression leads to altered binding of matrix molecules, and detrimental effects on cartilage integrity, is, however, still lacking. Paresh Jobanputra, DM, MRCP, MRCGP Hong Lin, Dip Med Karen Jenkins, BSc Charles Bavington, BSc Frank R. Brennan, PhD George Nuki, FRCP University of Edinburgh and Western General Hospital Donald M. Salter, MD, MRCPath Jane L. Godolphin, BSc University of Edinburgh Medical School Edinburgh, Scotland 1. Arner EC, Tortorella MD: Signal transduction through chondro- cyte integrin receptors induces matrix metalloproteinase synthesis and synergizcs with interleukin-1. Arthritis Rheum 38:1304-1314, 1995 2. Salter DM, Godolphin JL, Gourlay MS: Chondrocyte heterogcncity: variation of intcgrin expression at different sites in human fetal knccs. J Histochem Cytochem 43:447-457, 1995 3. Loeser KF: Modulation of intcgrin-mediated attachment of chondrocytes to extracellular matrix proteins by cations, retinoic acid, and transforming growth factor p. Exp Cell Res 21 1:17-23, 1994 CONCISE COMMUNICATIONS 1432 4. Salter DM, Hughes DB, Simpson R, Gardncr DL: Intcgrin expression by human articular chondrocytes. Br J Rheumatol 31:231-234, 1992 5. Guo J, Jourdian GW. MacCallum DK: Culture and growth characteristics of chondrocytes encapsulated in alginatc beads. Conn Tissue Rcs 19:277-297, 1989 6. Jobanputra P, Corrigall V, Kingsley G, Panayi G: Cellular rcsponses to human chondrocytes: absence of allogcncic responses in the presence of HLA-IIK and ICAM-1. Clin Exp Immunol 9c):336-344, 1992 7. Locscr RF, Carlson CS, McGcc MP: Expression of pl integrins by cultured articular chondrocytes and in osteoarthritic cartilage. Exp Cell Res 217:248-257, 1995 8. LaFlarnme SE, Akiyama S, Yamada K: Regulation of fibroncctin receptor distribution. J Cell Biol 117:437-447, 1992 9. Dalton SL, iMarcantonio EE, Associan KK: Cell attachment controls fibronectin and a5pl integrin levels in fibroblasts. J Biol Chem 267:8186-8191, 1992 10. Woods VL Jr, Schrcck PJ, Gcsink DS, Pacheco 110, Amiel 11. Akcson WH, Lotz M: Intcgrin expression by human articular chondrocytcs. Arthritis Rheum 37:537-544, 1994 Monocyte apoptosis in patients with active lupus The prcscncc of nucleosome polymers in the peripheral blood of patients with active systemic lupus erythematows (SLE) (1) suggests ongoing apoptosis. While the lymphocyte population has been implicated as a source of these fragments (2), thc cells that release the chromatin are incompletely characterized. Our group has reported that patients with active lupus have a T cell subsct which spontaneously lyses autologous macrophagcs in vitro (3), and that murine T cells capable of inducing apoptosis in syngcneic macrophages will cause a lupus-like disease in vivo (4). These results suggest that monocyteimacrophage apoptosis may occur in patients with active lupus. Wc therefore sought to determine whcthcr patients with active lupus have evidence of ongoing monocyte apop tosis. Patients with active lupus were recruited from the inpatient rheumatology servicc and outpatient clinics at the University of Michigan, and paticnts with inactive lupus and active rheumatoid arthritis (RA) wcrc recruited from the outpatient clinics. Hcalthy control subjccts were obtained by advertising. Paticnts met the American College of liheumatology (formerly, the Amcrican Rhcumatism Association) criteria for the classification of SLE ( 5 ) or RA (6). SLE and KA activity were defined as previously described (3). Patients receiving cyclophosphamide were excluded from study. Some samples were stored as heparinized whole blood overnight at 23°C. In these cases, care was taken to store both patient and control samples under identical conditions. All samples wcrc analyzed within 20 hours of phlebotomy. To detect apoptotic cells, peripheral blood mononuclear cells (PBMC) were isolated as described ( 3 ) , and stained with propidium iodidc (PI) and Hoescht 33342, and then analyzed by flow cytomctry (7,8). The lymphocyte, monocytc, and granulocyte populations were identified by forward and 90" light scatter and confirmed by staining with anti-CD3FITC, anti-B1 -FITC, and anti-MO1-FITC (Coulter, Ilialeah, FL). Dead cells were identified by high PI staining, viable cells by low PI staining and 2N IIoescht DNA staining, and apoptotic cells by low PI staining and less than 2N Hoescht D N A staining (7,8). Where indicated, T cells were depletcd by rosetting with sheep erythrocytes (3), and the nonrosetting population was fixed and analyzed by transmission clcctron microscopy (TEM) as described (4). Differences between groups were tested by analysis of variance, using SYSTAT software (Evanston, IL), or by t-tests. PBMC were obtained from 10 paticnts with activc SLE, 8 patients with inactive SLE, 7 patients with activc RA, and 12 healthy controls. The cells were stained as describcd above, and the percentages of apoptotic monocytes and lymphocytes were determined by flow cytometry. Figure 1A shows a representative histogram of monocytcs from a hcalthy control. Figurc 1B shows monocytcs from a patient with active lupus analyzed similarly. In Figure lB, the population of cclls excluding PI, and containing less D N A than controls by Hoescht staining, contains thc apoptotic cells (7,8). Figure 2A shows that an average of 5% of the mono- 6 A LogDNA - LogDNA - Figure 1. Peripheral blood mononuclear cells from A, a hcalthy control and R, a patient with activc lupus were stained with anti-MO1-FII'C, propidium iodidc, and Hocscht 33342 and analyzed using 3-color flow cytomctry. 'I'hc monocytc population is shown. The x axis (Log DNA) rcprcscnts the intcnsity of thc Hocscht staining; thc y axis (Log Viability) represents the PI staining intensity. The unlabelled z (vertical) axis rcprcscnts the cell nurnbcr.