1427
THEDEVELOPMENTOF MONOCLONAL ANTIBODIES TO THE HUMAN MITOCHONDRIAL 60-kd HEAT-SHOCK PROTEIN, AND THEIR USE IN STUDYING THE EXPRESSION O F THE PROTEIN IN RHEUMATOID ARTHRITIS MOHAMMED SHARIF, JENNIFER G. WORRALL, BHAG SINGH, RADHEY S. GUPTA, PETER M. LYDYARD, CLAUDE LAMBERT, JOHN McCULLOCH, and GRAHAM A. ROOK Objective. To assess the claim that the human 60-kd heat-shock protein (HSP) is highly expressed in the joints of patients with rheumatoid arthritis (RA), but is not readily detected in normal tissues. Methods. Monoclonal antibodies were raised against the human 60-kd mitochondria1 heat-shock protein (P1 protein; hsp60), and their specificity was established. They were then applied to synovial tissue. Results. HSP was expressed similarly in normal, osteoarthritic, and RA synovium. Low levels of hsp6O were detected in synovial fluid by immunoprecipitation. From the Departments of Medical Microbiology, Immunology, and Rheumatology, University College and Middlesex School of Medicine, London, United Kingdom; and the Department of Biochemistry, McMaster University, Hamilton, Ontario, Canada. Dr. Sharifs work was supported by a grant from the Arthritis and Rheumatism Council. Mohammed Sharif, PhD: Research Assistant, Department of Medical Microbiology, University College and Middlesex School of Medicine (current address: Rheumatology Unit, Bristol Royal Infirmary, Bristol, UK); Jennifer G. Worrall, MBBS: Senior Registrar, Bloomsbury Rheumatology Unit, University College and Middlesex School of Medicine; Bhag Singh, PhD: Research Associate, Department of Biochemistry, McMaster University; Radhey S. Gupta, PhD: Professor, Department of Biochemistry, McMaster University; Peter M. Lydyard, PhD: Reader, Department of Immunology, University College and Middlesex School of Medicine; Claude Lambert, MB: Research Assistant, Department of Medical Microbiology, University College and Middlesex School of Medicine (current address: Service de Nephrologie, CHU St. Etienne, Hopital Nord, St. Priest en Jarez, France); John McCulloch, PhD: Research Assistant, Department of Medical Microbiology, University College and Middlesex School of Medicine; Graham A. Rook, MD: Reader, Department of Medical Microbiology, University College and Middlesex School of Medicine. Address reprint requests to Graham A. Rook, MD, Department of Medical Microbiology, University College and Middlesex School of Medicine, 67-73 Riding House Street, London W l P 7PP, United Kingdom. Submitted for publication April 30, 1992; accepted in revised form August 19, 1992. Arthritis and Rheumatism, Vol. 35, No. 12 (December 1992)
Conclusion. Minor differences in the distribution of hsp60 in synovium from RA joints were attributable to increased cellularity and to the disorganization of the tissue architecture.
The strong evidence that the mycobacterial 65-kd heat-shock protein (hsp65) is involved in several rodent models of autoimmunity (1-4) led to speculation that it might also be involved in the pathogenesis of rheumatoid arthritis (RA) in humans. This hypothesis was strengthened by other types of evidence linking mycobacteria with RA, including changes in antibody and skin-test responses to crude mycobacterial antigen preparations, the presence of antibody to hsp65 ( 5 ) , the strikingly increased percentage of agalactosyl IgG seen in RA and the mycobacterioses, and the presence of arthritis in some patients with mycobacterioses (for review, see ref. 6). These facts led to the hypothesis (7) that exposure to cross-reactive bacterial HSPs causes the RA patients to develop autoreactive T cells which recognize the mitochondrial 60-kd heat-shock protein, the human homolog of hsp65 (8). This notion arose despite the fact that rodent hsp60 is not implicated in the rat model of arthritis. The human and mycobacterial proteins share 40% identity, with a further 20% of conservative changes, making such cross-reactive responses a possibility. Some authors have therefore sought evidence for RA-specific changes in the expression of hsp60 in the synovial tissue and synovial fluid (9). However, work in this area has been hampered by a lack of reagents specific for the human homolog. We report here the development of monoclonal antibodies (MAb) with specificity for the human hsp60 or PI protein (8), and show that in contrast to the
SHARIF ET AL
1428
claims of earlier studies, hsp60 is constitutively expressed in normal synovium, and changes in disease expression are not specific to RA9 but reflect the disorganization of the tissue.
MATERIALS AND METHODS Human hsp60 preparations. Four different preparations of human hsp60 (referred to as P1 protein in previous publications) were used. (i) Crude P1 protein purified from human placenta by ion-exchange and gel-exclusion chromatography (Singh B, Gupta RS: unpublished observations). In this preparation, 5 0 4 0 % of total protein was hsp60. (ii) Human placental protein purified by an affinity column bearing a polyclonal antibody raised against the hamster homolog. (iii) Two recombinant forms of the P1 protein PKK13A and PKK13D. PKK13A (PlA) contains most of the human hsp60 sequence except for the N-terminal30 amino acids, and it gives a single band at approximately 57 kd on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). In PKK13D, the N-terminal 168 amino acids are lacking, and on SDS-PAGE gels, it gives a single band of approximately 40 kd. Both these latter preparations were obtained by electroelution from SDS-PAGE gels, and their preparation and characterization are described elsewhere (10). Immunizationof BALB/c mice with human hsp60. On day 0 and day 5 , BALBlc mice were given intramuscular injections (at 4 sites) of 50 pg of the crude P1 preparation in 0.2 ml of Freund’s incomplete adjuvant. Seven weeks after the first injection, they received 20 pg of the affinity-purified P1 protein, intravenously, in 0.2 ml of Ribi adjuvant, prepared as recommended by the manufacturer (Ribi ImmunoChem, Hamilton, MT). Spleens were harvested for production of hybridomas 4 days after this injection. The mice were bled by cardiac puncture at the same time, and the serum was collected for testing on recombinant P1 protein by enzyme-linked immunosorbent assay (ELISA). Hybridomas, with JK(P3-X63-Ag8, 653) cells as fusion partners, were generated using standard procedures (11). Screening of culture supernatants by ELISA. Culture supernatants were initially screened by ELISA on the 2 recombinant forms of hsp60 (PKkl3A and PKK13D) and on a mixture of 2 negative control antigens (human IgG and albumin) to eliminate polyreactive IgM antibodies. Antigens were coated onto wells at 2 &ml, and ELISAs were performed as described previously (5). Promising supernatants were re-screened by immunoblotting on PKK13A and RA synovial fluid, and on the mixture of human IgG and human serum albumin (HSA). Hybridomas were then cloned by limiting dilution, and grown as ascites in pristaneprepared peritoneal cavities. Screening of the culture supernatants and ascites by Western blots. Supernatants from all the clones were screened using miniature Western blots of several HSPs and cell lysates, as well as by ELISA on Mycobucterium leprue and Escherichiu coli hsp65. Preparative miniature blots of mycobacterial hsp65, E coli hsp65, mycobacterial and human hsp70, and cell lysates from peripheral blood mononuclear cells (PBMC) and a Jurkat T cell line (JRT) were
prepared using standard protocols. Briefly, 10 pg of recombinant HSP, 150 cLg of RA synovial fluid, or 300 pg of cell lysates (derived from approximately 5 X 10‘ cells) were resolved by SDS-PAGE on slab gels (0.75-mm thick) of 10% (volumeholume) acrylamide, and transferred electrophoretically to nitrocellulose membrane in Tris-glycine-methanol buffer using a wet blotter (Mini Transphor TE 22; Hoefer, San Francisco, CA), at a constant 50V for 1 hour or at 1OV overnight, at room temperature. After blocking the nitrocellulose membrane with 1% bovine serum albumin/phosphate buffered saline (PBS)/ Tween 20, the blots were stained with the MAb, at 50 pg/ml, using a miniblotter (model MN 28; Biometra, Kent, UK) as recommended by the manufacturer. Binding of MAbs and the rabbit polyclonal antibody to P1 or to the recombinant preparations was detected using peroxidase-labeled rabbit antiserum to mouse immunoglobulin (DAKO P260; Dako, Burlingame, CA) or swine anti-rabbit immunoglobulin (DAKO P217). Epitope mapping by ELISA cross-inhibition assays and by binding to deletions of human hsp60. Antibodies were classified into groups by standard cross-inhibition in ELISAs, using wells coated with 1 &ml of P1A. They were also studied using a series of well-characterized deletions of human hsp60 expressed in E coli, as described elsewhere (12). The recombinant human protein from several of these deletions has been purified by chromatography and subsequent elution from SDS-PAGE. To examine the antibody reactivity, 10 pg of protein from various deletions, as well as mature P1 protein, were separated on SDS-PAGE, and after blotting onto nitrocellulose, they were separately reacted with dilutions of the different MAb. Immunoprecipitationof hsp60 from RA synovial fluid. To further check the specificity of the MAb, 2 immunoprecipitation protocols were used. The first was a standard fluid-phase protocol, using the monoclonal anti-hsp60 (2E1/53), which had been found to be the most efficient capturing antibody in a solid-phase assay. Precipitation was induced with rabbit anti-mouse IgM (lot M-8644; Sigma, St. Louis, MO). Alternatively, 2E1/53 was purified on a Sepharose gel’ flltration column (Sigma CL-6B-200) and Dynabeads (M-450) were coated with the antibody by direct adsorption. The beads were washed in sterile distilled water, and a uniform suspension was obtained by brief shaking. The purified MAb was dissolved in 0.2M borate buffer (pH 9.5) to a concentration of 150 pg/ml. Then, an equal volume of the suspension of Dynabeads was added to the antibody solution, producing an antibody-to-bead ratio of 75 mg:15 mg. After 24 hours of incubation, at room temperature by slow, end-over-end rotation, the Dynabeads were collected with a magnetic particle concentrator (MPC-1; Dynal, Great Neck, NY). The supernatants were discarded while the tube was in the MPC, and the coated beads were then washed sequentially, starting with 5 ml of 0.01M PBSiTween 20 for 1 hour, followed by a 12-hour wash with 5 ml of 0.01M sodium chloride, 0.01% merthiolate, 0.1% Tween 20 (PH 7.5). Finally, the coated beads were washed with PBS and resuspended in PBS to a concentration of approximately 4 x lo8 Dynabeads/ml(30 mg/ml). Coated Dynabeads were stored at 4°C with either merthiolate (0.01%) or sodium azide (0.02%) as a bacteriostatic agent.
EXPRESSION OF THE 60-kd HSP IN ARTHRITIS For immunoprecipitation, 5 pl of a suspension of the coated beads was mixed with 200 pl of synovial fluid in 2 ml of PBS. After 24 hours of incubation by slow, end-over-end rotation at 4"C, the Dynabeads were collected using the MPC as before. Dynabeads were washed several times with PBS, concentrated using the MPC, and subjected to SDSPAGE. The bound hsp60 was subsequently detected on Western blots using biotin-labeled IgG2a MAb, 4B9189. Immunocytochemistry and immunohistochemistry. HEp-2 cells were cultured on glass coverslips, washed free of culture medium with PBS, air-dried, and fixed in cold acetone for 10 minutes. Samples of rheumatoid (n = 5 ) and osteoarthritic (n = 3) synovium were obtained from patients undergoing prosthetic arthroplasty, and samples of normal synovium (n = 3) from patients undergoing amputation of limbs for proximal malignancy. Cryostat sections (7p) were air-dried and fixed in cold acetone for 10 minutes. All fixed samples were then rehydrated in PBS, incubated for 10 minutes with 10% normal goat serum to block nonspecific binding, and then for 30 minutes with the MAb 4B9189 at a concentration of 100 pglml in PBS. Samples were washed 3 times in PBS and then incubated for 30 minutes with fluorescein-conjugated goat anti-mouse IgG (Sera-Lab, Crawley Down, Sussex, UK), at 1 :20 in PBS, in the presence of 10% normal human serum. Samples were washed in PBS and mounted in glycerol containing the anti-fade agent 1-4-diazabicyclo-(2,2,2,)octane (DABCO; Sigma, Paisley, UK). Controls consisted of samples incubated with PBS or with irrelevant MAb of the same isotype instead of the specific MAb.
RESULTS Screening of the culture supernatants. We had anticipated that it would be difficult to raise antibodies to the human HSP in mice, since it differs from the rodent sequence by only 13 amino acids (12); however, this proved not to be so. Approximately 1.4 x lo7 spleen cells were plated in 384 wells. Ninety wells showed growth. Culture supernatants from all the growing wells were screened by ELISA and immunoblot. Fifty-two of the 90 supernatants bound to PKK13A by ELISA, but not to wells coated with HSA and human IgG. Of these, 34 bound to miniblots of PKK13A. Some of these gave spurious bands on immunoblots of RA synovial fluid, and were not studied further; however, 15 were specific by this criterion. Ten of these were re-cloned, established as ascites in pristane-prepared BALB/c mice, and subjected to further study. These ascites preparations all bound very strongly on immunoblots of both the full-length and the truncated recombinant forms of the human hsp60, even when used at a concentration of 1 pg/ml (blots not shown). All the hybridomas selected were IgM, except for clone 4B9/89, which was IgG2a.
1429
Screening of the MAbs on HSPs and cell lysates. All MAbs bound to a band of approximate molecular mass 58 kd on Western blots of cell lysates from PBMC and a Jurkat T cell line (Figure 1). The rabbit polyclonal antibody to the P1 protein also bound to this band. On blots of L929 cell lysates (murine cell line), 5 of the MAbs bound weakly to a band of 58 kd; however, the rabbit polyclonal antibody bound very strongly to this band. The monoclonals did not bind to any bands on Western blots of human hsp70 or mycobacterial hsp70, and only 2 of the monoclonals bound to mycobacterial hsp65 and E coli hsp65 (1A8/20 and 3G9/92). This binding was weak. Binding to deletions of hsp60 and findings of cross-inhibitionassays. Partial mapping of the epitopes bound by these antibodies was achieved by screening them on a series of deletions. This revealed that the antibodies fall into 3 groups. The results of analysis obtained from cross-inhibition assays on the PlA protein were broadly compatible with the pattern of binding on the P1 deletion peptides. This allowed the antibodies to be divided into 3 main groups; 1 could not be classified (3A4/9), and was inhibited only by itself. Thus, the epitopes recognized by 3 antibodies could be localized to amino acids 21 1-288. Four others bound to 288-366, and 2 bound either to 335-366 or to 484547. Cross-inhibition studies confirmed that 2E1/53 (epitope between 21 1-288) was, as anticipated, totally unable to block the IgG monoclonal 4B9/89 (epitope between 335-366 or 484-547). Neither of these bound to the M leprue or E coli hsp65. Pilot experiments subsequently revealed that this pair of antibodies can function as a simple capture assay, and for immunoprecipitation. Capture assay and immunoprecipitation of hsp60. In spite of the clear evidence for their specificity, none of the antibodies described here bound to immunoblots of RA synovial fluid under the conditions described. These results suggested that the hsp60, if present in RA synovial fluid, might be present at rather lower concentrations than previously thought. We therefore used capture assays and immunoprecipitation to further assess the specificity of the MAb, to verify the presence of the protein in RA synovial fluid, and to achieve a rough estimate of its concentration. The lower detection limit of the capture assay using recombinant PKK 13A as a standard was approximately 0.75 pg/ml. This was identical whether the standard was dissolved in PBS or in synovial fluid. By this technique, detectable quantities of the
SHARIF ET AL
1430
Figure 1. Immunoblots of cell lysates from peripheral blood mononuclear cells (PBMC) and the Jurkat T cell line (JRT) probed with monoclonal antibodies (MAb) 3C8/65 and 4B9/89. Lane 1, Molecular weight markers; lanes 2,3, and 4, cell lysates of PBMC stained with Aurodye, 3C8/65, and 4B9/89, respectively; lanes 5,6, and 7, cell lysates of JRT stained with Aurodye, 3C8/65, and 4B9/89, respectively; lane 8, PKK13A (PlA) stained with MAb 4B9/89 (positive control).
antigen appeared to be present in 1 of 3 RA synovial fluids tested. Since 2E1/53 is an IgM, cross-linking by rheumatoid factor is not likely to be a problem. However, the quantities detected were close to the detection limit of the assay (data not shown), and were not regarded as reliable. Therefore, in order to confirm the presence of the HSP, it was immunoprecipitated from RA synovial fluid using the 2 techniques described above. In conventional immunoprecipitates, those collected at 3,000 revolutions per minute contained most of the hsp60. Figure 2 shows a result obtained using Dynabeads. Immunocytochemistry and immunohistochemistry. High-power fluorescence microscopy of HEp-2 cells revealed that the pattern of MAb 4B9/89 staining was composed of multiple small, oval outlines distributed throughout the cytoplasm, including the peripheries of cytoplasmic processes (Figure 3A). All synovial tissues showed extensive staining with MAb 4B9/89. Normal synovium showed intense cytoplasmic staining of the synovial lining cells and fainter staining of the cells in the subintima and blood vessel walls (Figure 3B). Rheumatoid synovium showed widespread cytoplasmic staining of a variety of morphologic cell types (Figure 3C). Cells of the lining layer and within dense mononuclear cell aggregates showed faint-to-moderate staining; the most intense staining was seen on large mononuclear cells scattered singly throughout the deeper layers. Faint
Figure 2. 2E1/53, an IgM monoclonal antibody (MAb) to the human mitochondria1 60-kd heat-shock protein (hsp60), was attached to Dynabeads, and used to isolate hsp60 from a solution of PKK13A in phosphate buffered saline (PBS; positive control), and from rheumatoid arthritis (RA) synovial fluid. Beads were subjected to electrophoresis, and immunoblots were prepared from the detached proteins. Blots were probed with biotin-labeled 4B9/89 (IgG MAb to recombinant hsp60 (PKK13A) (additional hsp60). Lane 1, 4 positive control); lane 2, PKK13A isolated from PBS (10 pg in 200 PI); lane 3, hsp60 isolated from 200 pl of RA synovial fluid. (See Patients and Methods for further details.)
EXPRESSION OF THE 60-kd HSP IN ARTHRITIS
Figure 3. Indirect immunofluorescence studies with monoclonal antibody (MAb) 4B9/89. A, HEp-2 cell line stained with IgG MAb 4B9/89. The granular cytoplasmic pattern is characteristic of mitochondria1 staining. B, Normal human synovium showing binding of MAb 4B9/89 predominantly to cells of the lining layer. C, Rheumatoid synovium showing binding of MAb 4B9/89 distributed extensively throughout the tissue. D, Osteoarthritic synovium showing binding of MAb 4B9/89 distributed throughout the tissue, similar to that in rheumatoid synovium. (Magnification x 1,000 in A, and x 200 in B-D.)
1431
SHARIF ET AL
1432
patchy staining of blood vessel walls was also present. Osteoarthritic synovium with an inflammatory infiltrate showed a pattern of staining indistinguishable from that of rheumatoid synovium (Figure 3D); less inflamed specimens resembled normal synovium. Controls (not shown) demonstrated no photographable fluorescence.
DISCUSSION We report here the production and use of MAb binding to the human homolog of the mycobacterial hsp65. We found that BALB/c mice readily produce such antibodies when immunized with the human protein, and surprisingly, the epitopes recognized, while not fully defined, are mostly confined to the mammalian HSP. Cross-reactive binding to the mycobacterial or E coli proteins was rare and weak. We have proved the specificity of our antibodies by immunoblotting and by immunoprecipitation. Moreover, monoclonal 439189 applied to fixed, permeabilized HEp-2 cells identified cytoplasmic bodies with the typical size, shape, and distribution of mitochondria; this is consistent with the described cellular location of hsp60 (8). Immunohistochemical studies using MAb 4B9/89 showed hsp60 to be widely distributed in rheumatoid synovium, where it was expressed in a variety of cell types. A similar distribution was found in inflamed osteoarthritic synovium, indicating that expression of hsp60 was not disease-specific. Furthermore, the presence of hsp60 in normal synovium, most notably in the cytoplasm of lining cells, confirmed that the molecule is constitutively expressed by normal cells. This was confirmed by gel electrophoresis of lysed normal PBMC, and is consistent with the known chaperonin role of this protein. The authors of a recent study of synovium from juvenile chronic arthritis (JCA) concluded that the expression of hsp60 is actually increased in these patients (13). It is conceivable that this patient group differs from adult RA, and it is in the JCA group that the most convincing evidence of recognition of the human protein has been obtained (14). Karlsson-Parra et a1 (9), in immunohistochemical studies using MAb ML30, raised against the mycobacterial homolog of human hsp60 but claimed to cross-react with the human form, found intense staining of rheumatoid tissues but no staining of either normal tissues or of a variety of nonrheumatoid disease control tissues. A separate study (15) using MAb ML30 showed staining of several bands on Western
blots of lysed rat synovium, suggesting lack of specificity. Interestingly, the same study demonstrated immunohistochemical staining with ML30 of synovium from normal rats and from rats with adjuvant arthritis and collagen-induced arthritis. These results, and ours, cast doubt on the reliability of ML30 as a probe for hsp60. ML30 was not screened for use on human tissues, but rather, was selected as specific for the mycobacterial hsp65 on extracts of these organisms. On human (and rat) tissues, it binds to several components; an immunoblot illustrating this point has been published by other investigators (15). Similarly, ML30 appears to cross-react with a cell surface component in rodents, also recognized by a polyclonal antibody to the mycobacterial hsp65 (16), but the significance or identity of this epitope is not yet clear. Our own preliminary flow cytometry studies with F(ab’), fragments of 4B9/89 suggest that hsp60 is not expressed on cell membranes (Smith M, Yuksel F, McCulloch J, Rook GA, Lydyard P: unpublished observations). Our results confirm the ubiquity of hsp60 but provide no evidence for a specific etiologic role in rheumatoid arthritis. The data on responses of T cells to this antigen remain disparate, and recognition of HSP peptides presented by class I1 major histocompatibility complex molecules may be a normal mechanism for the removal of stressed or transformed cells, which is present in all inflammatory sites.
REFERENCES 1. Van Eden W, Thole JER, van der Zee R, Noordzij A,
van Embden JDA, Hensen EJ, Cohen IR: Cloning of the mycobacterial epitope recognised by T lymphocytes in adjuvant arthritis. Nature 331: 171-173, 1988 2. Van den Broek MF, van Bruggen MCJ, Hogervorst EJM, van Eden W, van der Zee R, van den Berg WB: Protection against streptococcal cell wall-induced arthritis by pretreatment with the mycobacterial 65kDa heat shock protein. J Exp Med 170M9-466, 1989 3. Billingham MEJ, Carney S, Butler R, Colston MJ: A mycobacterial65kDa heat shock protein induces antigenspecific suppression of adjuvant arthritis, but is not itself arthritogenic. J Exp Med 171:33%344, 1990 4. Thompson SJ, Rook GAW, Brealey R, van der Zee R, Elson CJ: Autoimmune reactions to heat shock proteins in pristane-induced arthritis. Eur J Immunol 20:247% 2484, 1990 5. Bahr GM, Rook GAW, Al-Saffar M, van Embden JDA,
Stanford JL, Behbehani K: Antibody levels to mycobacteria in relation to HLA type: evidence for non-HLAlinked high levels of antibody to the 65kDa heat shock
EXPRESSION OF THE 60-kd HSP IN ARTHRITIS
6. 7.
8.
9.
10.
11.
protein of M. tuberculosis in rheumatoid arthritis. Clin Exp Immunol74:211-215, 1988 Rook GAW, Stanford JL: Slow bacterial infections or autoimmunity. Immunol Today 13:160-164, 1992 Lamb JR, Bal V, Mendez-Sempeiro P, Mehlert A, So A, Rothbard J, Jindal J, Young RA, Young DB: Stress proteins may provide a link between the immune response to infection and autoimmunity. Int Immunol 1: 191-196, 1989 Jindal S, Dudani AK, Singh B, Harley CB, Gupta RS: Primary structure of a human mitochondrial protein homologous to the bacterial and plant chaperonins and to the 65 kilodalton mycobacterial antigen. Mol Cell Biol 9:2279-2283, 1989 Karlsson-Parra A, Soderstrom K, Ferm M, Ivanyi J, Kiessling R, Klareskog L: Presence of human 65kDa heat shock protein (HSP) in inflamed joints and subcutaneous nodules of RA patients. Scand J Immunol 31:283-288, 1990 Singh B, Gupta RS: Expression of human 60kDa heat shock proteins (HSP60 or P1) in Escherichia coli and development and characterization of corresponding monoclonal antibodies. DNA Cell Biol 11:48W96, 1992 Rook GAW, Steele J, Rademacher T: A monoclonal antibody raised by immunising mice with group A strep-
12. 13.
14.
15.
16.
1433
tococci binds to agalactosyl IgG from rheumatoid arthritis. Ann Rheum Dis 47:247-250, 1988 Venner TJ, Gupta RS: Nucleotide sequence of mouse hsp60 (chaperonin, GroEL homolog) cDNA. Biochim Biophys Acta 1087:336-338, 1990 Boog CJP, de Graeff-Meeder ER, Lucassen MA, van der Zee R, Voorhorst-Ogink MM, van Kooten PJS, Geuze HJ, van Eden W: Two monoclonal antibodies generated against human hsp60 show reactivity with synovial membranes of patients with juvenile chronic arthritis. J Exp Med 175:1805-1810, 1992 De Graeff-Meeder ER, van der Zee R, Rijkers GT, Schuurman HJ, Kuis W, Bijlsma JW, Zegers BJ, van Eden W: Recognition of human 60 kD heat shock protein by mononuclear cells from patients with juvenile chronic arthritis. Lancet 337:1368-1372, 1991 Kleinau S, Soderstrom K, Kiessling R, Klareskog L: A monoclonal antibody to the mycobacterial 65 kDa heat shock protein (ML 30) binds to cells in normal and arthritic joints of rats. Scand J Immunol 33:195-202, 1991 Wand-Wurttenburger A, Schoel B, Ivanyi J, Kaufman SH: Surface expression by mononuclear phagocytes of an epitope shared with mycobacterial heat shock protein 60. Eur J Immunol21:108!&1092, 1991
THEDEVELOPMENTOF MONOCLONAL ANTIBODIES TO THE HUMAN MITOCHONDRIAL 60-kd HEAT-SHOCK PROTEIN, AND THEIR USE IN STUDYING THE EXPRESSION O F THE PROTEIN IN RHEUMATOID ARTHRITIS MOHAMMED SHARIF, JENNIFER G. WORRALL, BHAG SINGH, RADHEY S. GUPTA, PETER M. LYDYARD, CLAUDE LAMBERT, JOHN McCULLOCH, and GRAHAM A. ROOK Objective. To assess the claim that the human 60-kd heat-shock protein (HSP) is highly expressed in the joints of patients with rheumatoid arthritis (RA), but is not readily detected in normal tissues. Methods. Monoclonal antibodies were raised against the human 60-kd mitochondria1 heat-shock protein (P1 protein; hsp60), and their specificity was established. They were then applied to synovial tissue. Results. HSP was expressed similarly in normal, osteoarthritic, and RA synovium. Low levels of hsp6O were detected in synovial fluid by immunoprecipitation. From the Departments of Medical Microbiology, Immunology, and Rheumatology, University College and Middlesex School of Medicine, London, United Kingdom; and the Department of Biochemistry, McMaster University, Hamilton, Ontario, Canada. Dr. Sharifs work was supported by a grant from the Arthritis and Rheumatism Council. Mohammed Sharif, PhD: Research Assistant, Department of Medical Microbiology, University College and Middlesex School of Medicine (current address: Rheumatology Unit, Bristol Royal Infirmary, Bristol, UK); Jennifer G. Worrall, MBBS: Senior Registrar, Bloomsbury Rheumatology Unit, University College and Middlesex School of Medicine; Bhag Singh, PhD: Research Associate, Department of Biochemistry, McMaster University; Radhey S. Gupta, PhD: Professor, Department of Biochemistry, McMaster University; Peter M. Lydyard, PhD: Reader, Department of Immunology, University College and Middlesex School of Medicine; Claude Lambert, MB: Research Assistant, Department of Medical Microbiology, University College and Middlesex School of Medicine (current address: Service de Nephrologie, CHU St. Etienne, Hopital Nord, St. Priest en Jarez, France); John McCulloch, PhD: Research Assistant, Department of Medical Microbiology, University College and Middlesex School of Medicine; Graham A. Rook, MD: Reader, Department of Medical Microbiology, University College and Middlesex School of Medicine. Address reprint requests to Graham A. Rook, MD, Department of Medical Microbiology, University College and Middlesex School of Medicine, 67-73 Riding House Street, London W l P 7PP, United Kingdom. Submitted for publication April 30, 1992; accepted in revised form August 19, 1992. Arthritis and Rheumatism, Vol. 35, No. 12 (December 1992)
Conclusion. Minor differences in the distribution of hsp60 in synovium from RA joints were attributable to increased cellularity and to the disorganization of the tissue architecture.
The strong evidence that the mycobacterial 65-kd heat-shock protein (hsp65) is involved in several rodent models of autoimmunity (1-4) led to speculation that it might also be involved in the pathogenesis of rheumatoid arthritis (RA) in humans. This hypothesis was strengthened by other types of evidence linking mycobacteria with RA, including changes in antibody and skin-test responses to crude mycobacterial antigen preparations, the presence of antibody to hsp65 ( 5 ) , the strikingly increased percentage of agalactosyl IgG seen in RA and the mycobacterioses, and the presence of arthritis in some patients with mycobacterioses (for review, see ref. 6). These facts led to the hypothesis (7) that exposure to cross-reactive bacterial HSPs causes the RA patients to develop autoreactive T cells which recognize the mitochondrial 60-kd heat-shock protein, the human homolog of hsp65 (8). This notion arose despite the fact that rodent hsp60 is not implicated in the rat model of arthritis. The human and mycobacterial proteins share 40% identity, with a further 20% of conservative changes, making such cross-reactive responses a possibility. Some authors have therefore sought evidence for RA-specific changes in the expression of hsp60 in the synovial tissue and synovial fluid (9). However, work in this area has been hampered by a lack of reagents specific for the human homolog. We report here the development of monoclonal antibodies (MAb) with specificity for the human hsp60 or PI protein (8), and show that in contrast to the
SHARIF ET AL
1428
claims of earlier studies, hsp60 is constitutively expressed in normal synovium, and changes in disease expression are not specific to RA9 but reflect the disorganization of the tissue.
MATERIALS AND METHODS Human hsp60 preparations. Four different preparations of human hsp60 (referred to as P1 protein in previous publications) were used. (i) Crude P1 protein purified from human placenta by ion-exchange and gel-exclusion chromatography (Singh B, Gupta RS: unpublished observations). In this preparation, 5 0 4 0 % of total protein was hsp60. (ii) Human placental protein purified by an affinity column bearing a polyclonal antibody raised against the hamster homolog. (iii) Two recombinant forms of the P1 protein PKK13A and PKK13D. PKK13A (PlA) contains most of the human hsp60 sequence except for the N-terminal30 amino acids, and it gives a single band at approximately 57 kd on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). In PKK13D, the N-terminal 168 amino acids are lacking, and on SDS-PAGE gels, it gives a single band of approximately 40 kd. Both these latter preparations were obtained by electroelution from SDS-PAGE gels, and their preparation and characterization are described elsewhere (10). Immunizationof BALB/c mice with human hsp60. On day 0 and day 5 , BALBlc mice were given intramuscular injections (at 4 sites) of 50 pg of the crude P1 preparation in 0.2 ml of Freund’s incomplete adjuvant. Seven weeks after the first injection, they received 20 pg of the affinity-purified P1 protein, intravenously, in 0.2 ml of Ribi adjuvant, prepared as recommended by the manufacturer (Ribi ImmunoChem, Hamilton, MT). Spleens were harvested for production of hybridomas 4 days after this injection. The mice were bled by cardiac puncture at the same time, and the serum was collected for testing on recombinant P1 protein by enzyme-linked immunosorbent assay (ELISA). Hybridomas, with JK(P3-X63-Ag8, 653) cells as fusion partners, were generated using standard procedures (11). Screening of culture supernatants by ELISA. Culture supernatants were initially screened by ELISA on the 2 recombinant forms of hsp60 (PKkl3A and PKK13D) and on a mixture of 2 negative control antigens (human IgG and albumin) to eliminate polyreactive IgM antibodies. Antigens were coated onto wells at 2 &ml, and ELISAs were performed as described previously (5). Promising supernatants were re-screened by immunoblotting on PKK13A and RA synovial fluid, and on the mixture of human IgG and human serum albumin (HSA). Hybridomas were then cloned by limiting dilution, and grown as ascites in pristaneprepared peritoneal cavities. Screening of the culture supernatants and ascites by Western blots. Supernatants from all the clones were screened using miniature Western blots of several HSPs and cell lysates, as well as by ELISA on Mycobucterium leprue and Escherichiu coli hsp65. Preparative miniature blots of mycobacterial hsp65, E coli hsp65, mycobacterial and human hsp70, and cell lysates from peripheral blood mononuclear cells (PBMC) and a Jurkat T cell line (JRT) were
prepared using standard protocols. Briefly, 10 pg of recombinant HSP, 150 cLg of RA synovial fluid, or 300 pg of cell lysates (derived from approximately 5 X 10‘ cells) were resolved by SDS-PAGE on slab gels (0.75-mm thick) of 10% (volumeholume) acrylamide, and transferred electrophoretically to nitrocellulose membrane in Tris-glycine-methanol buffer using a wet blotter (Mini Transphor TE 22; Hoefer, San Francisco, CA), at a constant 50V for 1 hour or at 1OV overnight, at room temperature. After blocking the nitrocellulose membrane with 1% bovine serum albumin/phosphate buffered saline (PBS)/ Tween 20, the blots were stained with the MAb, at 50 pg/ml, using a miniblotter (model MN 28; Biometra, Kent, UK) as recommended by the manufacturer. Binding of MAbs and the rabbit polyclonal antibody to P1 or to the recombinant preparations was detected using peroxidase-labeled rabbit antiserum to mouse immunoglobulin (DAKO P260; Dako, Burlingame, CA) or swine anti-rabbit immunoglobulin (DAKO P217). Epitope mapping by ELISA cross-inhibition assays and by binding to deletions of human hsp60. Antibodies were classified into groups by standard cross-inhibition in ELISAs, using wells coated with 1 &ml of P1A. They were also studied using a series of well-characterized deletions of human hsp60 expressed in E coli, as described elsewhere (12). The recombinant human protein from several of these deletions has been purified by chromatography and subsequent elution from SDS-PAGE. To examine the antibody reactivity, 10 pg of protein from various deletions, as well as mature P1 protein, were separated on SDS-PAGE, and after blotting onto nitrocellulose, they were separately reacted with dilutions of the different MAb. Immunoprecipitationof hsp60 from RA synovial fluid. To further check the specificity of the MAb, 2 immunoprecipitation protocols were used. The first was a standard fluid-phase protocol, using the monoclonal anti-hsp60 (2E1/53), which had been found to be the most efficient capturing antibody in a solid-phase assay. Precipitation was induced with rabbit anti-mouse IgM (lot M-8644; Sigma, St. Louis, MO). Alternatively, 2E1/53 was purified on a Sepharose gel’ flltration column (Sigma CL-6B-200) and Dynabeads (M-450) were coated with the antibody by direct adsorption. The beads were washed in sterile distilled water, and a uniform suspension was obtained by brief shaking. The purified MAb was dissolved in 0.2M borate buffer (pH 9.5) to a concentration of 150 pg/ml. Then, an equal volume of the suspension of Dynabeads was added to the antibody solution, producing an antibody-to-bead ratio of 75 mg:15 mg. After 24 hours of incubation, at room temperature by slow, end-over-end rotation, the Dynabeads were collected with a magnetic particle concentrator (MPC-1; Dynal, Great Neck, NY). The supernatants were discarded while the tube was in the MPC, and the coated beads were then washed sequentially, starting with 5 ml of 0.01M PBSiTween 20 for 1 hour, followed by a 12-hour wash with 5 ml of 0.01M sodium chloride, 0.01% merthiolate, 0.1% Tween 20 (PH 7.5). Finally, the coated beads were washed with PBS and resuspended in PBS to a concentration of approximately 4 x lo8 Dynabeads/ml(30 mg/ml). Coated Dynabeads were stored at 4°C with either merthiolate (0.01%) or sodium azide (0.02%) as a bacteriostatic agent.
EXPRESSION OF THE 60-kd HSP IN ARTHRITIS For immunoprecipitation, 5 pl of a suspension of the coated beads was mixed with 200 pl of synovial fluid in 2 ml of PBS. After 24 hours of incubation by slow, end-over-end rotation at 4"C, the Dynabeads were collected using the MPC as before. Dynabeads were washed several times with PBS, concentrated using the MPC, and subjected to SDSPAGE. The bound hsp60 was subsequently detected on Western blots using biotin-labeled IgG2a MAb, 4B9189. Immunocytochemistry and immunohistochemistry. HEp-2 cells were cultured on glass coverslips, washed free of culture medium with PBS, air-dried, and fixed in cold acetone for 10 minutes. Samples of rheumatoid (n = 5 ) and osteoarthritic (n = 3) synovium were obtained from patients undergoing prosthetic arthroplasty, and samples of normal synovium (n = 3) from patients undergoing amputation of limbs for proximal malignancy. Cryostat sections (7p) were air-dried and fixed in cold acetone for 10 minutes. All fixed samples were then rehydrated in PBS, incubated for 10 minutes with 10% normal goat serum to block nonspecific binding, and then for 30 minutes with the MAb 4B9189 at a concentration of 100 pglml in PBS. Samples were washed 3 times in PBS and then incubated for 30 minutes with fluorescein-conjugated goat anti-mouse IgG (Sera-Lab, Crawley Down, Sussex, UK), at 1 :20 in PBS, in the presence of 10% normal human serum. Samples were washed in PBS and mounted in glycerol containing the anti-fade agent 1-4-diazabicyclo-(2,2,2,)octane (DABCO; Sigma, Paisley, UK). Controls consisted of samples incubated with PBS or with irrelevant MAb of the same isotype instead of the specific MAb.
RESULTS Screening of the culture supernatants. We had anticipated that it would be difficult to raise antibodies to the human HSP in mice, since it differs from the rodent sequence by only 13 amino acids (12); however, this proved not to be so. Approximately 1.4 x lo7 spleen cells were plated in 384 wells. Ninety wells showed growth. Culture supernatants from all the growing wells were screened by ELISA and immunoblot. Fifty-two of the 90 supernatants bound to PKK13A by ELISA, but not to wells coated with HSA and human IgG. Of these, 34 bound to miniblots of PKK13A. Some of these gave spurious bands on immunoblots of RA synovial fluid, and were not studied further; however, 15 were specific by this criterion. Ten of these were re-cloned, established as ascites in pristane-prepared BALB/c mice, and subjected to further study. These ascites preparations all bound very strongly on immunoblots of both the full-length and the truncated recombinant forms of the human hsp60, even when used at a concentration of 1 pg/ml (blots not shown). All the hybridomas selected were IgM, except for clone 4B9/89, which was IgG2a.
1429
Screening of the MAbs on HSPs and cell lysates. All MAbs bound to a band of approximate molecular mass 58 kd on Western blots of cell lysates from PBMC and a Jurkat T cell line (Figure 1). The rabbit polyclonal antibody to the P1 protein also bound to this band. On blots of L929 cell lysates (murine cell line), 5 of the MAbs bound weakly to a band of 58 kd; however, the rabbit polyclonal antibody bound very strongly to this band. The monoclonals did not bind to any bands on Western blots of human hsp70 or mycobacterial hsp70, and only 2 of the monoclonals bound to mycobacterial hsp65 and E coli hsp65 (1A8/20 and 3G9/92). This binding was weak. Binding to deletions of hsp60 and findings of cross-inhibitionassays. Partial mapping of the epitopes bound by these antibodies was achieved by screening them on a series of deletions. This revealed that the antibodies fall into 3 groups. The results of analysis obtained from cross-inhibition assays on the PlA protein were broadly compatible with the pattern of binding on the P1 deletion peptides. This allowed the antibodies to be divided into 3 main groups; 1 could not be classified (3A4/9), and was inhibited only by itself. Thus, the epitopes recognized by 3 antibodies could be localized to amino acids 21 1-288. Four others bound to 288-366, and 2 bound either to 335-366 or to 484547. Cross-inhibition studies confirmed that 2E1/53 (epitope between 21 1-288) was, as anticipated, totally unable to block the IgG monoclonal 4B9/89 (epitope between 335-366 or 484-547). Neither of these bound to the M leprue or E coli hsp65. Pilot experiments subsequently revealed that this pair of antibodies can function as a simple capture assay, and for immunoprecipitation. Capture assay and immunoprecipitation of hsp60. In spite of the clear evidence for their specificity, none of the antibodies described here bound to immunoblots of RA synovial fluid under the conditions described. These results suggested that the hsp60, if present in RA synovial fluid, might be present at rather lower concentrations than previously thought. We therefore used capture assays and immunoprecipitation to further assess the specificity of the MAb, to verify the presence of the protein in RA synovial fluid, and to achieve a rough estimate of its concentration. The lower detection limit of the capture assay using recombinant PKK 13A as a standard was approximately 0.75 pg/ml. This was identical whether the standard was dissolved in PBS or in synovial fluid. By this technique, detectable quantities of the
SHARIF ET AL
1430
Figure 1. Immunoblots of cell lysates from peripheral blood mononuclear cells (PBMC) and the Jurkat T cell line (JRT) probed with monoclonal antibodies (MAb) 3C8/65 and 4B9/89. Lane 1, Molecular weight markers; lanes 2,3, and 4, cell lysates of PBMC stained with Aurodye, 3C8/65, and 4B9/89, respectively; lanes 5,6, and 7, cell lysates of JRT stained with Aurodye, 3C8/65, and 4B9/89, respectively; lane 8, PKK13A (PlA) stained with MAb 4B9/89 (positive control).
antigen appeared to be present in 1 of 3 RA synovial fluids tested. Since 2E1/53 is an IgM, cross-linking by rheumatoid factor is not likely to be a problem. However, the quantities detected were close to the detection limit of the assay (data not shown), and were not regarded as reliable. Therefore, in order to confirm the presence of the HSP, it was immunoprecipitated from RA synovial fluid using the 2 techniques described above. In conventional immunoprecipitates, those collected at 3,000 revolutions per minute contained most of the hsp60. Figure 2 shows a result obtained using Dynabeads. Immunocytochemistry and immunohistochemistry. High-power fluorescence microscopy of HEp-2 cells revealed that the pattern of MAb 4B9/89 staining was composed of multiple small, oval outlines distributed throughout the cytoplasm, including the peripheries of cytoplasmic processes (Figure 3A). All synovial tissues showed extensive staining with MAb 4B9/89. Normal synovium showed intense cytoplasmic staining of the synovial lining cells and fainter staining of the cells in the subintima and blood vessel walls (Figure 3B). Rheumatoid synovium showed widespread cytoplasmic staining of a variety of morphologic cell types (Figure 3C). Cells of the lining layer and within dense mononuclear cell aggregates showed faint-to-moderate staining; the most intense staining was seen on large mononuclear cells scattered singly throughout the deeper layers. Faint
Figure 2. 2E1/53, an IgM monoclonal antibody (MAb) to the human mitochondria1 60-kd heat-shock protein (hsp60), was attached to Dynabeads, and used to isolate hsp60 from a solution of PKK13A in phosphate buffered saline (PBS; positive control), and from rheumatoid arthritis (RA) synovial fluid. Beads were subjected to electrophoresis, and immunoblots were prepared from the detached proteins. Blots were probed with biotin-labeled 4B9/89 (IgG MAb to recombinant hsp60 (PKK13A) (additional hsp60). Lane 1, 4 positive control); lane 2, PKK13A isolated from PBS (10 pg in 200 PI); lane 3, hsp60 isolated from 200 pl of RA synovial fluid. (See Patients and Methods for further details.)
EXPRESSION OF THE 60-kd HSP IN ARTHRITIS
Figure 3. Indirect immunofluorescence studies with monoclonal antibody (MAb) 4B9/89. A, HEp-2 cell line stained with IgG MAb 4B9/89. The granular cytoplasmic pattern is characteristic of mitochondria1 staining. B, Normal human synovium showing binding of MAb 4B9/89 predominantly to cells of the lining layer. C, Rheumatoid synovium showing binding of MAb 4B9/89 distributed extensively throughout the tissue. D, Osteoarthritic synovium showing binding of MAb 4B9/89 distributed throughout the tissue, similar to that in rheumatoid synovium. (Magnification x 1,000 in A, and x 200 in B-D.)
1431
SHARIF ET AL
1432
patchy staining of blood vessel walls was also present. Osteoarthritic synovium with an inflammatory infiltrate showed a pattern of staining indistinguishable from that of rheumatoid synovium (Figure 3D); less inflamed specimens resembled normal synovium. Controls (not shown) demonstrated no photographable fluorescence.
DISCUSSION We report here the production and use of MAb binding to the human homolog of the mycobacterial hsp65. We found that BALB/c mice readily produce such antibodies when immunized with the human protein, and surprisingly, the epitopes recognized, while not fully defined, are mostly confined to the mammalian HSP. Cross-reactive binding to the mycobacterial or E coli proteins was rare and weak. We have proved the specificity of our antibodies by immunoblotting and by immunoprecipitation. Moreover, monoclonal 439189 applied to fixed, permeabilized HEp-2 cells identified cytoplasmic bodies with the typical size, shape, and distribution of mitochondria; this is consistent with the described cellular location of hsp60 (8). Immunohistochemical studies using MAb 4B9/89 showed hsp60 to be widely distributed in rheumatoid synovium, where it was expressed in a variety of cell types. A similar distribution was found in inflamed osteoarthritic synovium, indicating that expression of hsp60 was not disease-specific. Furthermore, the presence of hsp60 in normal synovium, most notably in the cytoplasm of lining cells, confirmed that the molecule is constitutively expressed by normal cells. This was confirmed by gel electrophoresis of lysed normal PBMC, and is consistent with the known chaperonin role of this protein. The authors of a recent study of synovium from juvenile chronic arthritis (JCA) concluded that the expression of hsp60 is actually increased in these patients (13). It is conceivable that this patient group differs from adult RA, and it is in the JCA group that the most convincing evidence of recognition of the human protein has been obtained (14). Karlsson-Parra et a1 (9), in immunohistochemical studies using MAb ML30, raised against the mycobacterial homolog of human hsp60 but claimed to cross-react with the human form, found intense staining of rheumatoid tissues but no staining of either normal tissues or of a variety of nonrheumatoid disease control tissues. A separate study (15) using MAb ML30 showed staining of several bands on Western
blots of lysed rat synovium, suggesting lack of specificity. Interestingly, the same study demonstrated immunohistochemical staining with ML30 of synovium from normal rats and from rats with adjuvant arthritis and collagen-induced arthritis. These results, and ours, cast doubt on the reliability of ML30 as a probe for hsp60. ML30 was not screened for use on human tissues, but rather, was selected as specific for the mycobacterial hsp65 on extracts of these organisms. On human (and rat) tissues, it binds to several components; an immunoblot illustrating this point has been published by other investigators (15). Similarly, ML30 appears to cross-react with a cell surface component in rodents, also recognized by a polyclonal antibody to the mycobacterial hsp65 (16), but the significance or identity of this epitope is not yet clear. Our own preliminary flow cytometry studies with F(ab’), fragments of 4B9/89 suggest that hsp60 is not expressed on cell membranes (Smith M, Yuksel F, McCulloch J, Rook GA, Lydyard P: unpublished observations). Our results confirm the ubiquity of hsp60 but provide no evidence for a specific etiologic role in rheumatoid arthritis. The data on responses of T cells to this antigen remain disparate, and recognition of HSP peptides presented by class I1 major histocompatibility complex molecules may be a normal mechanism for the removal of stressed or transformed cells, which is present in all inflammatory sites.
REFERENCES 1. Van Eden W, Thole JER, van der Zee R, Noordzij A,
van Embden JDA, Hensen EJ, Cohen IR: Cloning of the mycobacterial epitope recognised by T lymphocytes in adjuvant arthritis. Nature 331: 171-173, 1988 2. Van den Broek MF, van Bruggen MCJ, Hogervorst EJM, van Eden W, van der Zee R, van den Berg WB: Protection against streptococcal cell wall-induced arthritis by pretreatment with the mycobacterial 65kDa heat shock protein. J Exp Med 170M9-466, 1989 3. Billingham MEJ, Carney S, Butler R, Colston MJ: A mycobacterial65kDa heat shock protein induces antigenspecific suppression of adjuvant arthritis, but is not itself arthritogenic. J Exp Med 171:33%344, 1990 4. Thompson SJ, Rook GAW, Brealey R, van der Zee R, Elson CJ: Autoimmune reactions to heat shock proteins in pristane-induced arthritis. Eur J Immunol 20:247% 2484, 1990 5. Bahr GM, Rook GAW, Al-Saffar M, van Embden JDA,
Stanford JL, Behbehani K: Antibody levels to mycobacteria in relation to HLA type: evidence for non-HLAlinked high levels of antibody to the 65kDa heat shock
EXPRESSION OF THE 60-kd HSP IN ARTHRITIS
6. 7.
8.
9.
10.
11.
protein of M. tuberculosis in rheumatoid arthritis. Clin Exp Immunol74:211-215, 1988 Rook GAW, Stanford JL: Slow bacterial infections or autoimmunity. Immunol Today 13:160-164, 1992 Lamb JR, Bal V, Mendez-Sempeiro P, Mehlert A, So A, Rothbard J, Jindal J, Young RA, Young DB: Stress proteins may provide a link between the immune response to infection and autoimmunity. Int Immunol 1: 191-196, 1989 Jindal S, Dudani AK, Singh B, Harley CB, Gupta RS: Primary structure of a human mitochondrial protein homologous to the bacterial and plant chaperonins and to the 65 kilodalton mycobacterial antigen. Mol Cell Biol 9:2279-2283, 1989 Karlsson-Parra A, Soderstrom K, Ferm M, Ivanyi J, Kiessling R, Klareskog L: Presence of human 65kDa heat shock protein (HSP) in inflamed joints and subcutaneous nodules of RA patients. Scand J Immunol 31:283-288, 1990 Singh B, Gupta RS: Expression of human 60kDa heat shock proteins (HSP60 or P1) in Escherichia coli and development and characterization of corresponding monoclonal antibodies. DNA Cell Biol 11:48W96, 1992 Rook GAW, Steele J, Rademacher T: A monoclonal antibody raised by immunising mice with group A strep-
12. 13.
14.
15.
16.
1433
tococci binds to agalactosyl IgG from rheumatoid arthritis. Ann Rheum Dis 47:247-250, 1988 Venner TJ, Gupta RS: Nucleotide sequence of mouse hsp60 (chaperonin, GroEL homolog) cDNA. Biochim Biophys Acta 1087:336-338, 1990 Boog CJP, de Graeff-Meeder ER, Lucassen MA, van der Zee R, Voorhorst-Ogink MM, van Kooten PJS, Geuze HJ, van Eden W: Two monoclonal antibodies generated against human hsp60 show reactivity with synovial membranes of patients with juvenile chronic arthritis. J Exp Med 175:1805-1810, 1992 De Graeff-Meeder ER, van der Zee R, Rijkers GT, Schuurman HJ, Kuis W, Bijlsma JW, Zegers BJ, van Eden W: Recognition of human 60 kD heat shock protein by mononuclear cells from patients with juvenile chronic arthritis. Lancet 337:1368-1372, 1991 Kleinau S, Soderstrom K, Kiessling R, Klareskog L: A monoclonal antibody to the mycobacterial 65 kDa heat shock protein (ML 30) binds to cells in normal and arthritic joints of rats. Scand J Immunol 33:195-202, 1991 Wand-Wurttenburger A, Schoel B, Ivanyi J, Kaufman SH: Surface expression by mononuclear phagocytes of an epitope shared with mycobacterial heat shock protein 60. Eur J Immunol21:108!&1092, 1991
