Eur. J. Immunol. 1991.21: 1297-1302
Pieter C. M. Res., Daniela L. M. Orsini., Jacob M.Van Laar., Anneke A. M. Janson., Christiane Abou-ZeidA and R e d R. I? DeVries.
M. tuberculosis-reactive synovial fluid T cell clones
Diversity in antigen recognition by Mycobacterium tuberculosis-reactiveT cell clones from the synovial fluid of rheumatoid arthritis patients
Department of lmmunobematology and Blood Bank, University Hospital., Department of Rheumatology, University Hospital., Leiden and Department of Medical Microbiology, University College and Middlesex School of Medicine., London
In a previous study we have shown that synovial fluid mononuclear cells from many rheumatoid arthritis (RA) patients exhibit an enhanced response to M. tuberculosis antigens as compared to peripheral blood mononuclear cells. The 65-kDa heat-shock protein of M.tuberculosis was shown not to play an important role in this response, therefore other mycobacterial proteins must be involved. In this study we have investigated the possibility that synovial fluid Tcells from RA patients predominantly recognize a limited number of M. tuberculosis antigens, as a result of a lesion-specific activation of only those M. tuberculosis-reactive T cells that have cross-reacted with joint-related autoantigens. From the synovial fluid of four RA patients M. tuberculosis-reactiveTcell clones were isolated and analyzed for their phenotype, HLA-DR restriction and proliferation to immunoblot fractions containing sodium dodecyl sulfate-polyacrylamide gel-separated M. tuberculosis proteins of known molecular weight range. The overall M. tuberculosis immunoblot recognition pattern of the clones was strikingly heterogeneous. Within a panel of 15 clones 12 different antigenic specificities could be distinguished. In other words,we did not observe a dominant recognition of a few M. tuberculosis antigens by synovial fluid T cells. This argues against the hypothesis that the elevated synovial Tcell reactivity against M.tuberculosis is a reflection of an in vivo expansion of a limited number of different types of M. tuberculosis-reactive T cells as a result of a cross-reaction with putative joint autoantigens.
1 Introduction The etiology of rheumatoid arthritis (RA) is unknown, but a considerable interest in a possible involvement of T cells cross-reactive with M. tuberculosis antigens and self antigens of the joint has developed during recent years.TheseT cells are thought to play a role in the initiation and perpetuation of the inflammation process within the affected joints. The hypothesis originated from the adjuvant arthritis (AA) model in Lewis rats in which the disease-inducing T cells respond to the 65-kDa heat shock protein of M. tuberculosis and cartilage proteoglycans [ 1, 21. The observation, that synovial fluid mononuclear cells (SFMC) of RA patients displayed an increased proliferative response to M.tuberculosis antigens compared to cells from the peripheral blood, was interpreted as being in favor of a direct involvement of M.tuberculosis-reactive Tcells in the autoimmune response to self antigens [3, 41. Notably, the enhanced responses to M. tuberculosis of synovial fluid (SF) Tcells might be a reflection of a selective expansion within the joint of those M.tuberculosis-reactiveTcellsthat have cross-reacted with joint components. A high frequency of these M. tuberculosis-reactive T cells within the SF would be the result from such a lesion-specific expansion. Although in the A A model the 65-kDa-reactive T cell clones have been isolated from LN of affected animals, an obvious link of RA with this model would be responses of
[I 91841 Correspondence: Pieter C. M. Res, Department of Immunohematology and Blood Bank, Bldg 1, E3-0, Postbus 9600, 2300 RC Leiden, The Netherlands Abbreviations: S F Synovial fluid
1297
RA: Rheumatoid arthritis
0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1991
SFMC against the 65-kDa protein. However, we found that a highly purified rBCG 65-kDa protein, that is identical with the M. tuberculosis 65-kDa protein, was not a major mycobacterial Ag recognized by SF Tcells [5]. This result was in contrast to data published by other groups [6,7] and our earlier data using a less purified rBCG 65-kDa preparation [4]. SF cell reactivity against this latter preparation appeared to be mainly directed to E. coli [ 5 ] . In this study we therefore investigated the possibility that the strong responses to M.tuberculosis are the outcome of a selective expansion within the joint of Tcells reactive with mycobacterial proteins other than the 65-kDa protein after a cross-reaction with joint components. Such components should be joint specific and would in all probability not include autologous heat-shock proteins, which contain many T cell epitopes also present on M. tuberculosis heat-shock proteins, but do not display a tissue distribution that would explain the site specificity of RA. Although an infection with M. tuberculosis probably offers hundreds of potential Tcell epitopes to the immune system, most likely only a few T cell epitope sequences on M. tuberculosis proteins will have enough molecular mimicry with sequences on joint autoantigens to induce T cell crossreactivity. Thus this model would imply a preferential recognition of a limited number of mycobacterial proteins/epitopes by SF Tcells, like for instance a 47-kDa M. tuberculosis containing immunoblot fraction, which has been described as a dominant M. tuberculosis Ag for DR4-positive PBMC from RA patients and normal controls [S].We have isolated a total of 30 M.tuberculosis-reactiveT cell clones from the SFof four RA patients and tested the in vitro proliferative response of 15 clones to immunoblot fractions containing SDS-PAGE-separated M.tuberculosis proteins of known molecular mass range. 0014-2980/9l/OSOS- 1297$3.SO + .25/0
1298
Eur. J. Immunol. 1991. 21: 1297-1302
P. C. M. Res, D. L. M. Orsini, J. M.Van Laar et al.
2 Material and methods
2.4 Prolieration assays
2.1 Patients
Cloned Tcells (10“) and 5 x 10“irradiated (2000 rad) PBMC as a source of APC were cultured in 96-well flat-bottom microtiter plates in 200 pl culture medium for 3 days with or without the addition of 20 pl of an M. tuberculosis immunoblot particle suspension or 10 pg/ml of a given Ag preparation. Culture medium consisted of IMDM with 10% type AB normal human serum. During the last 18 h the cells were cultured in the presence of tritiated thymidine, harvested and the incorporation was measured in a liquid scintillation counter.
Patients fulfilled the American Rheumatism Association criteria for classical or definite RA. Serological and cellular typing for determination of class I and class I1 HLA phenotypes was performed on PBMC of the patients. 2.2 Antigens and mitogens The M . tuberculosis ultrasonicate preparation was kindly donated by Paul Klatser from the Royal Tropical Institute and generated as follows. M. tuberculosis strain H 37 RA was ultrasonicated in 0.1 M PBS, pH 7.2, on ice for 16 min at 90 W and 50 duty cycles. The secreted 30/31-kDa proteins corresponding to the antigen 85 complex of BCG were purified from the culture filtrate of M. tuberculosis as previously described [9]. PHA was obtained from the Wellcome Foundation (Beckenham, GB).
2.5 Gel electrophoresis and immunoblotting
The M . tuberculosis ultrasonicate preparation (0.8 mg, 2.4 mg or 8 mg) was applied in a comb of 12 cm length on a 1.55 mm thick 11%SDS-PAGE gel and run for 10 cm under reducing conditions. Samples were obtained by dissolving the proteins in sample buffer and heating them for 5 min at 100°C. Sample buffer consisted of 62.5 mM Tris-HC1 pH 6.8, 2% SDS, 10% glycerol, 0.002% bromophenolblue and 5% 2-ME. After electrophoresis Ag were electrotrans2.3 Generation of M. tuberculosis-reactiveT cell clones ferred to nitrocellulose paper with a 0.2 pm pore size paper T cell clones were isolated from T cell lines that had been (Schleicher & Schuell, Dassel, FRG). Marker proteins generated by stimulation of SFMC of RA patients with 10 were also run on the gel and blotted to a particular part of pglml M. tuberculosis for 5 days and expansion with 10% the nitrocellulose filter which was stained with amido black Lymphocult (Biotest, Dreiech, FRG) for another 7 days. for molecular mass determinations. The nitrocellulose filter Clones from patients W, K and L were generated as follows. was cut into 22 25 x 4 m m strips and each strip was Cells were cultured at a concentration of 0.3 celVwel1 in converted into a microparticulate form as described [lo]; 96-round-bottom-well plates in 100 pl of culture medium the material was solubilized and sterilized in 500 pI DMSO consisting of IMDM with 10% type AB normal human for 1 h at room temperature. Precipitation was performed serum supplemented with 0.1 pg PHA, 10% Lymphocult, 2 by adding 500 pI 0.05 M sodium carbonatelhydrogen carx 10“ allogeneic 3000 rad-irradiated allogeneic PBMC from bonate buffer, pH 9.6 while shaking continuously. Each at least three different individuals and 2 x 103 allogeneic suspension was washed three times with RPMI, resusirradiated (5000 rad) EBV-transformed B cells (EBVB). pended in 1 ml RPMI and kept at -20°C. In a first After 3 days 100 p1 of culture medium supplemented with screening assay clones were tested with immunoblot fractions from the gel loaded with 2.4 mg M . tuberculosis. Most 10% Lymphocult was added per well and after another 4-7 days, cells from positive wells were transferred to macro- clones responded to one or a few adjacent fractions of the wells, restimulated with culture medium containing per ml “2.4 mg panel“. However, a few clones did not react with 1 pg PHA, 10% Lymphocult, lohirradiated PBMC and 105 any fraction of this panel; these were found to be positive irradiated EBVB and expanded in culture medium plus with fractions of the “10 mg panel“. Other clones re10% Lymphocult. Only the Tcell clones of patient H were sponded to all fractions of the “2.4 mg panel“, probably generated by means of an Ag-specific method. Instead of because of some overspill during electrophoresis Ag was PHA and allogeneic feeder cells a M . tuberculosis ultrason- present in all fractions.These clones showed a nice clear-cut icate suspension and autologous peripheral blood cells as a reaction pattern with the fractions of the “0.8 mg source of APC were used in this cloning procedure. panel”.
Table 1. M. fuberculosis-reactive T cell clones from SF of RA patientsa)
Patients
Cloning
M. tuberculosis-reactive Tcell clones
method
DR(Dw) typing
W DRl,DR4(Dw 14) L DRl,DR4(Dwl4) K DRl .DR4(Dw4) H DR4(Dw4.Dw14)
Stimulus
number
restriction
PHA PHA PHA
n = 6 n = 5 n = 14
DR1 (2) or Dw14 (4) DRl (1) or Dw14 (4) DRI (4) or Dw4 (10)
Antigen specific
I1 =
5
Dw4 (1) Dw4 and Dw14 (2) Dw14 (2)
4 M. tuberculosis-reactive Tcell clones were generated
from M. tuberculosis specific SF Tcell lines by LD either in the presence of PHA and allogeneic feeder cells or M. tuberculosis and autologous APC.
Eur. J. Imrnunol. 1991. 21: 1297-1302
M. tuberculosis-reactive synovial fluid T cell clones
1299
SDS-PAGE immunoblot particle suspensions. In this way the molecular mass of the protein(s) recognized by each clone could be assigned to a known range. Of the original 30 clones only 15 were tested. The other 15 either lost their specificity during restimulation cycles or were lost due to infection. As demonstrated in Fig. 1 for clone H54 the majority of clones responded specifically to one immunoblotfraction with low background reactivity to the other fractions, whereas some clones were positive with more than one adjacent immunoblotfraction.
3 Results 3.1 Isolation and restriction analysis of M.tuberculosis-reactive T cell clones
As shown in Table 1, M. tuberculosis-reactive Tcell clones were isolated from M.tuberculosis induced SF Tcell lines of four RA patients, whose SFMC showed a strong proliferative response in the presence of a M. tuberculosis ultrasonicate preparation (data not shown). T cells were cultured under LD conditions either in an Ag-dependent (patient H) or in a mitogen-dependent manner (patients W,K,L). A summary of the immunoblot analysis of the clones is Contrary to expectation, all clones isolated fromTcell lines shown inTable 3.Within the small series of clon,es from each of the patients W,K, and L were M. tuberculosis reactive, patient already many different M. tuberculosis Ag were whereas only 50% of the clones obtained from patient H recognized. Five clones reacted with the immunoblot responded to M. tuberculosis. All four individuals were fraction 12 containing proteins in the molecular mass range D(R) heterozygous and three out of four were DR lDR 4 . of 31-34-kDa. Because these clones recognized proteins in Most individual clones were restricted either via one or the this molecular mass range, we tried to distinguish them by other autologous HLA-DR determinant, but both types of testing their response against the major secreted antigen 85 clones were obtained from each individual. Examples of all complex of M. tuberculosis. This complex is seen as a different proliferationhestriction patterns of the clones that prominent doublet of 30 and 31-kDa on Western blots and we have observed are given in Table2. None of the consists of three closely related components 85A (identical DR1-restricted clones was restricted via DR4 and vice to P32), 85B (identical to a antigen) and 85C with extensive versa. The DR4 restriction was further analyzed by using Dw4+ and Dw14+ APC. Most clones proliferated to M. tuberculosis only when the Ag was presented by accessory cells carrying donor-matched HLA-Dw(4 or 14) subtype determinants (KC8). Some clones also slightly responded when the Ag was presented by DR4-matched but Dw(4 or 14)mismatched determinants (KB8 and W13). Two clones 81 from individual H (43 and 54) responded to the same extent CI whether Dw4+ or Dw14+ APC were used. Like most clones, the T cell lines from which the clones originated showed a much higher reactivity using APC with donormatched DR4 subtypes (data not shown). 21 0
3.2 T cell clone recognition of M. tuberculosis immunoblotfractions
IIlrnmL .mlr Imlr. C
1 2 3 4
t
94kDa
To investigate whether SF Tcells from RA patients display a preferential recognition of certain proteins and whether a relationship with DR1 and/or DR4 restriction can be found, the in vitro proliferative response of the Tcell clones was tested in the presence of matched APC and M. tuberculosis
I . I
5 6 7 8 9 10111213141518171819202122
t
67
t
43
t
30
t
20.1
t
14.4
Figure I. Example of the immunoblot fraction recognition pattern of a M . tuberculosis-reactiveTcellclone.Tcell clone cells (104) and 5 x 104 irradiated PBMC were cultured for 3 days in the presence of 20 pl of an immunoblot particle suspension. C = control fraction containing unloaded nitrocellulose.
Table 2. Restriction patterns of M . tuberculosis-reactiveT cell clonesa)
Clone
M. tubemulosir
APC DR1
KB8 KC8
+-
-
138 355 93 122 53 40613 453
+
nt
+ -
W13
+
w4
+
H54
230
-
-
DR1 Dw4 Dw4 Dw14 427 52037 61183 1193 258 217 117 190 267 14677 19150 352 122 97 140 217 172 763 30627 52 145 127 143 57750 nt nt 135 220 62 nt
nt
383
36153
1w
Dw14
3465
197 90 68
58807 140 467 75 nt
a) Tcells (104) and 5 x 104 DW4+ or Dw14+ irradiated (2000 rad) PBMC as APC were cultured for 3 days in the presence of 10 pg/ml M . tuberculosis.
1300
Eur. J. Immunol. 1991. 21: 1297-1302
P. C. M. Res, D. L. M. Orsini, J. M.Van Laar et al.
Table 3. Immunoblot analysis of M. tuberculosis-reactive T cell clonesa) Clonc
Restriction determinant
W4 w5 w13
12 21 +22 12
WY
DRl DR 1 Dw14 Dw14 Dwl4
LF7 LF5 LG4
Dw14 Dwl.1 Dw14
6 8 20
H38 H32 H13 HS4
Dwll
5 9 + 10 12 12 15
w1s
H16
12 17-20
Dwll
Dw4IDwlJ Dw4IDw 14 Dw4
KBX KC 1
Immunoblot fraction Corresponding recognized molecular mass (kDa)
14 14
DwJ
Dw4
31-34 14- 16 31-34 31-34 16-23 54-60
44-48 16-17 60-71 37-44 31-34 31-34 25-27 27-29 27-29
overlap in antibody recognition and N-terminal amino acid sequence [9, 11, 121. Another option to discriminate the fraction 12 reactive clones was a possible cross-reactivity with M. leprue. As shown inTable 4, the five clones showed three different Ag reactivity patterns on the basis of restriction via different DR molecules and the presence or absence of a proliferative response to M. leprue or to the M. tuberculosis 85 complex. Combining these data with the immunoblot data we conclude that within this series of 15 clones already 12 different Ag specificities can be distinguished. Thus the overall recognition pattern of M. tuberculosis proteins by these DR1- and DRCrestricted synovial T cell clones is quite heterogeneous.
4 Discussion The data presented in this study do not offer evidence for a preferential recognition of a limited number of M. tuberculosis proteins by SF T cells from RA patients. Within a panel of 15 DR1- and DRCrestricted M. tubercufosisreactive T cell clones isolated from the SF of four RA patients already 12 different Ag specificities were present. In this small panel of clones also no relationship between Ag recognition and DR1 and/or DR4 was found. The chance for a M. tuberculosis protein to contain a T cell epitope(s) that is also present on a joint-specific self Ag is
a) Tcells (104) and 5 X 104 irradiated PBMC were cultured for 3 days in the presence of 20 p1of a given immunoblot particle suspension. Three different panels of immunoblot fractions originating from gels loaded with respectively 0.8,2.4 or 8 mg M.tuberculosbwere available for this analysis. In most cases clones were positive with one or a few fractions of the “2.4 mg” panel, but some additional tests with one of the other panels were necessary to define the fraction(s) recognized (see Sect. 2.5).
probably small, therefore the cross-reactivity concept would predict a dominant recognition of only a few M. tuberculosis proteins by synovial T cells. The observed heterogeneity in Ag reactivity of the T cell clones argues against a selective expansion within the joint of only those types of M. tuberculosis-reactive T cells that have crossreacted with joint components. A heterogeneity in immunoblot fraction recognition was also found for nonlesional M. tubercufosis-reactive T cell clones from the peripheral blood of normal individuals and leprosy patients (A. Janson et al., manuscript in preparation). Only 1 out of 13 DRCrestricted T cell clones investigated in our study responded to the immunoblotfraction containing proteins in the 47-kDa range; an M. tuberculosis immunoblot fraction with proteins in this range was shown to be a major M. tuberculosis fraction recognized by peripheral blood Tcells from DR4+ RA patients and controls [8]. None of our TcR/a/P+ T cell clones responded to the immunoblotfraction containing M. tuberculosis proteins with a molecular mass < 10-kDa; this confirms the observation of Pfeffer et al. who demonstrated that only TcR y/6+ Tcells recognize column-separated M. tuberculosis components with an Mr < 10-kDa [13]. In an earlier study we have shown that responses to E. coli were found with similar frequency and magnitude in SFMC as responses to M. tuberculosis [5]. Usually the same samples
Table 4. Diversity in antigen specificity of Tcell clones responding t o M. tuberculosis immunoblot fraction 12 (31-34-kDa)a) (‘lcmc
W4 W13
w15 H.13 HS1
APC
DRI Dwll Dwll Dwl Dwl
No antigen 51
s2 67 383 137
M. tuberculosis
21 31 1 37 977 5 977 22 537
Antigen M. leprae
38 123 52 11650 4 788
M . tuberculosis 85 complex 63 18370 3648 nt 143
a) Tcells (104) and 5 x 104 irradiated PBMC as APC were grown for 3 days in the presence or absence of 10 &ml of a given Ag preparation.
Eur. J. Immunol. 1991. 21: 1297-1302
were positive for both Ag preparations. Therefore, we considered the possibility that at least part of the responses to M. tuberculosis and E. coli are directed to shared epitopes, perhaps expressed by homologous antigens like heat-shock proteins. However, none of the M.tuberculosisreactive clones presented in this report cross-reacts with E. coli (data not shown).Thus the E. coli reactivity of SFMC, which is just as prominent as the reactivity to M. tuberculosis, is directed to other antigens and thus exerted by different synovial T cells. In the same study we found responses to both M. tuberculosis and E. coli to be also present in a high percentage of pleural exudate mononuclear cell samples from non-RA patients with a chronic pleuritis not due tuberculosis. Enhanced Ag reactivity is therefore presumably a general characteristic of mononuclear cells from chronic inflammation sites. The M.tuberculosis-reactive T cell clones were specifically restricted via one HLA-DR or Dw determinant. Thus the Ag recognized by these T cell clones is probably not a superantigen since T cell responses against superantigens, although dependent on HLA class I1 molecules, are not specifically restricted by products of certain alleles [14, 151. The clones described in this study were restricted via DR1 or DR4. DR1 and especially the DR4 subtypes Dw4, Dw14 and Dw15 are restriction determinants that confer susceptibility for I U [16,17].DRl, Dw14 and Dw15 all share the amino acid (aa) 67-74 stretch on the third hypervariable region of the p chain of the HLA-DR molecule, whereas Dw4 only differs at position 71. These molecules differ strongly at this part of the p chain from other DR types and also from the DR4 subtype DwlO, which are not associated with a higher risk to develop RA. According to the so-called shared epitope hypothesis, aa 67-74 of the third hypervariable region of the p chain of DR1, Dw4, Dw14 and Dw15 molecules are causally related to susceptibility to RA [18]. Being part of the binding site for Ag these aa will be directly involved in TcR-Ag interactions critical for the development of RA, presumably as restriction sites for T cells responding to putative autoantigens. The shared epitope hypothesis most likely implies the autoreactive T cells to be restricted via all D(R) molecules that predispose for the development of RA and/or to recognize a similar peptide [MI. Since our clones, being either DR1 or DR4 restricted and probably recognizing different peptides, do not fulfil either of these criteria, it is not likely that the M. tuberculosis-reactive T cells isolated were directly involved in the response to self Ag. However, we do not necessarily have to conclude from our data that M. tuberculosis-reactive T cells do not play any role in the pathogenesis of RA. At least three explanations compatible with this notion are left. First, the in vitro properties of Tcells from the SF may not reflect the properties of cells from the synovium, for instance the CD4/CD8 ratio is usually much lower in the SF than in the synovium. It could be that those M. tuberculosis-reactive T cells that can cross-react to autoantigens are suppressed within the SF, whereas suppression of the same Tcells within the synovium and/or the peripheral blood for some reason is less effective.This would be analogous to the observation that Tcell clones responding to the myelin basic protein can be isolated from the peripheral blood but not from the cerebrospinal fluid of multiple sclerosis patients
M . tuberculosis-reactive synovial fluid T cell clones
1301
[19] and Haffler, personal communication). If the suppression of the disease causing SF Tcells would be mediated by epitopes on M. tuberculosis proteins different from those recognized by Th cells, then the presentation of small peptides containing only helper epitopes instead of whole bacteria might reveal otherwise obscured T cell specificities. Paradoxically, the 65-kDa heat-shock protein is an obvious candidate for inducing suppression of DR4restricted Tcells, because thus far none of the described 27 Tcell epitopes on the 65-kDa protein has been found to be DR4 restricted. Of course, none of the DRCrestricted T cell clones described in this article responds to the 65-kDa heat-shock protein either (data not shown). A second possibility is that M. tuberculosis-reactive SF T cells are directly stimulated by M. tuberculosis antigens present in the joint. In this case a cross-reactivity concept is not necessary, because the joint destruction will be caused by interleukins secreted by activated M. tuberculosisreactiveTcells. M.tuberculosis proteins diffused away from the site of infection might end up in the joint for instance because they bind specifically to joint components. The M. tuberculosis 85 complex is particulary interesting in this respect since it contains at least two proteins with the capability to bind to fibronectin, an extracellular matrix protein abudantly present in joints. The Tcell clones W13 and W15 described in this study are responsive to the M. tuberculosis 85 complex. Finally, a possible role for M. tuberculosis-reactiveTcells in the pathogenesis of RA may be that these cells do not react to autoantigens themselves, but give other Tcells the opportunity to do so. A cellular immune response induced by an infection with M.tuberculosis, which probably shares many homologous aa stretches with self proteins, may somehow interfere with preexisting networks involved in the control of undesired response to self antigens. This might give Tcells specific for endogenous antigens, in this case joint components, the chance to escape from regulation [20]. Perhaps not the T cells responding to self antigens but the M. tuberculosis-reactive Tcells which interact with the regulatory networks could be restricted by D(R) molecules that confer susceptibillity to RA.The M.tuberculosis-reactiveTcel1population with this capacity does not necessarily have to display a limited Ag recognition.
The authors wish to thank Tom Ottenhoff for helpful advice; Paul Klatser of the Royal Tropical Institute in Amsterdamfor performing the SDS-PAGE gel electrophoresis and subsequent blotting; Annemarie Termijtelenforproviding Tcell clones for typing the Dw type of DR4-positive cells; Ieke Schreuder and colleagues for serological HLA typing, and Irun Cohen and Jelle Thole for stimulating discussions. Received January 10, 1991.
5 References 1 Van Eden,W., Holoshitz, J., Nevo, Z., Frenkel, A., Klajman, A. and Cohen, I. R.,Proc. Nail. Acad. Sci. USA 1985. 82: 5117. 2 Van Eden, W.,Thole, J. E . R.,Van der Zee, R . , Noordzij, A., Van Embden, J. D. A., Hensen, E . J. and Cohen, I. R., Nature 1988. 331: 171.
1302
P. C. M. Res, D. L. M.Orsini, J. M.Van Laar et al.
3 Holoshitz, J., Klajman, A., Drucker, I., Lapidot, Z.,Yaretzky, A., Frenkel, A.,Van Eden,W. and Cohen, I. R., Lancet 1986. ii: 305. 4 Res, P. C. M., Schaar, C. G., Breedveld, F. C.,Van Eden,W.,Van Embden, J. D. A., Cohen, I. R. and DeVries, R. R. P., Lancet 1988. ii: 477. 5 Res, P. C. M., Telgt, D.,Van Laar, J. M., Oudkerk Pool, M., Breedveld, F. C. and De Vries, R. R. P., Lancet 1990. 336: 1406. 6 Gaston, J. S. H., Life, P. F., Bailey, L. C. and Bacon, I? A., J. Immunol. 1989. 143: 2494. 7 Soderstrom, K., Halapi, E., Nilsson, E., Gronberg, A., Van Embden, J. D. A., Klareskog, L. and Kiessling, R., Scand. J. Immunol. 1990. 32: 503. 8 Palacios-Boix, A. A., Estrada, G. I., Colston, M. J. and Panayi, G. S., J. Immunol. 1988. 140: 1844. 9 Abou-Zeid, C., Ratliff, T. L., Wiker, H. G., Harboe, M., Bennedsen, J. and Rook, G. A. W., Infect. Immun. 1988. 56: 3046. 10 Abou-Zeid, C., Filley, E., Steele, J. and Rook, G. A. W., J. Immunol. Methods 1987. 85: 5.
Eur. J. Immunol. 1991. 21: 1297-1302
11 De Wit, L., De la Cuvellerie, A., Ooms, J. and Content, J., Nucleic Acids Rex 1990. 18: 3995. 12 Wiker, H. G., Sletten, K., Nagai, S. and Harboe, M., Infect. Immun. 1990. 58: 272. 13 Pfeffer, K., Schoel, B., Gulle, H., Kaufmann, S. H. E. and Wagner, H., Eur. J. Immunol. 1990. 20: 1175. 14 Fleischer, B., Schrezenmeier, H. and Conradt, I?, Cell. Immunol. 1989. 120: 92. 15 Dellabona, P., Peccoud, J., Kappler, J., Marrack, F!, Benoist, C. and Mathis, D., Cell 1990. 62: 1115. 16 Wordsworth, B. P., Lanchbury, J. S. S., Sakkas, L. I.,Welsh, K. I., Panayi, G. S. and Bell, J. I., Proc. Narl. Acad. Sci. USA 1989. 86: 10049. 17 Watanabe,Y. ,Tokunaga, K., Matsuki, K. ,Takeuchi, F., Matsuta, K., Maeda, H., Omoto, K. and Takeo, J. J., J. Exp. Med. 1989. 169: 2263. 18 Gregersen, P. K., Silver, J. and Winchester, R. J., Arthritis Rheum. 1987. 30: 1205. 19 Ota, K., Matsui, M., Milford, E. L., Mackin, G. A.,Weiner, H. L. and Haffler, D. A., Nature 1990. 346: 183. 20 Cohen, I. R. and Young, D. B., Immunol. Today, in press.
Pieter C. M. Res., Daniela L. M. Orsini., Jacob M.Van Laar., Anneke A. M. Janson., Christiane Abou-ZeidA and R e d R. I? DeVries.
M. tuberculosis-reactive synovial fluid T cell clones
Diversity in antigen recognition by Mycobacterium tuberculosis-reactiveT cell clones from the synovial fluid of rheumatoid arthritis patients
Department of lmmunobematology and Blood Bank, University Hospital., Department of Rheumatology, University Hospital., Leiden and Department of Medical Microbiology, University College and Middlesex School of Medicine., London
In a previous study we have shown that synovial fluid mononuclear cells from many rheumatoid arthritis (RA) patients exhibit an enhanced response to M. tuberculosis antigens as compared to peripheral blood mononuclear cells. The 65-kDa heat-shock protein of M.tuberculosis was shown not to play an important role in this response, therefore other mycobacterial proteins must be involved. In this study we have investigated the possibility that synovial fluid Tcells from RA patients predominantly recognize a limited number of M. tuberculosis antigens, as a result of a lesion-specific activation of only those M. tuberculosis-reactive T cells that have cross-reacted with joint-related autoantigens. From the synovial fluid of four RA patients M. tuberculosis-reactiveTcell clones were isolated and analyzed for their phenotype, HLA-DR restriction and proliferation to immunoblot fractions containing sodium dodecyl sulfate-polyacrylamide gel-separated M. tuberculosis proteins of known molecular weight range. The overall M. tuberculosis immunoblot recognition pattern of the clones was strikingly heterogeneous. Within a panel of 15 clones 12 different antigenic specificities could be distinguished. In other words,we did not observe a dominant recognition of a few M. tuberculosis antigens by synovial fluid T cells. This argues against the hypothesis that the elevated synovial Tcell reactivity against M.tuberculosis is a reflection of an in vivo expansion of a limited number of different types of M. tuberculosis-reactive T cells as a result of a cross-reaction with putative joint autoantigens.
1 Introduction The etiology of rheumatoid arthritis (RA) is unknown, but a considerable interest in a possible involvement of T cells cross-reactive with M. tuberculosis antigens and self antigens of the joint has developed during recent years.TheseT cells are thought to play a role in the initiation and perpetuation of the inflammation process within the affected joints. The hypothesis originated from the adjuvant arthritis (AA) model in Lewis rats in which the disease-inducing T cells respond to the 65-kDa heat shock protein of M. tuberculosis and cartilage proteoglycans [ 1, 21. The observation, that synovial fluid mononuclear cells (SFMC) of RA patients displayed an increased proliferative response to M.tuberculosis antigens compared to cells from the peripheral blood, was interpreted as being in favor of a direct involvement of M.tuberculosis-reactive Tcells in the autoimmune response to self antigens [3, 41. Notably, the enhanced responses to M. tuberculosis of synovial fluid (SF) Tcells might be a reflection of a selective expansion within the joint of those M.tuberculosis-reactiveTcellsthat have cross-reacted with joint components. A high frequency of these M. tuberculosis-reactive T cells within the SF would be the result from such a lesion-specific expansion. Although in the A A model the 65-kDa-reactive T cell clones have been isolated from LN of affected animals, an obvious link of RA with this model would be responses of
[I 91841 Correspondence: Pieter C. M. Res, Department of Immunohematology and Blood Bank, Bldg 1, E3-0, Postbus 9600, 2300 RC Leiden, The Netherlands Abbreviations: S F Synovial fluid
1297
RA: Rheumatoid arthritis
0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1991
SFMC against the 65-kDa protein. However, we found that a highly purified rBCG 65-kDa protein, that is identical with the M. tuberculosis 65-kDa protein, was not a major mycobacterial Ag recognized by SF Tcells [5]. This result was in contrast to data published by other groups [6,7] and our earlier data using a less purified rBCG 65-kDa preparation [4]. SF cell reactivity against this latter preparation appeared to be mainly directed to E. coli [ 5 ] . In this study we therefore investigated the possibility that the strong responses to M.tuberculosis are the outcome of a selective expansion within the joint of Tcells reactive with mycobacterial proteins other than the 65-kDa protein after a cross-reaction with joint components. Such components should be joint specific and would in all probability not include autologous heat-shock proteins, which contain many T cell epitopes also present on M. tuberculosis heat-shock proteins, but do not display a tissue distribution that would explain the site specificity of RA. Although an infection with M. tuberculosis probably offers hundreds of potential Tcell epitopes to the immune system, most likely only a few T cell epitope sequences on M. tuberculosis proteins will have enough molecular mimicry with sequences on joint autoantigens to induce T cell crossreactivity. Thus this model would imply a preferential recognition of a limited number of mycobacterial proteins/epitopes by SF Tcells, like for instance a 47-kDa M. tuberculosis containing immunoblot fraction, which has been described as a dominant M. tuberculosis Ag for DR4-positive PBMC from RA patients and normal controls [S].We have isolated a total of 30 M.tuberculosis-reactiveT cell clones from the SFof four RA patients and tested the in vitro proliferative response of 15 clones to immunoblot fractions containing SDS-PAGE-separated M.tuberculosis proteins of known molecular mass range. 0014-2980/9l/OSOS- 1297$3.SO + .25/0
1298
Eur. J. Immunol. 1991. 21: 1297-1302
P. C. M. Res, D. L. M. Orsini, J. M.Van Laar et al.
2 Material and methods
2.4 Prolieration assays
2.1 Patients
Cloned Tcells (10“) and 5 x 10“irradiated (2000 rad) PBMC as a source of APC were cultured in 96-well flat-bottom microtiter plates in 200 pl culture medium for 3 days with or without the addition of 20 pl of an M. tuberculosis immunoblot particle suspension or 10 pg/ml of a given Ag preparation. Culture medium consisted of IMDM with 10% type AB normal human serum. During the last 18 h the cells were cultured in the presence of tritiated thymidine, harvested and the incorporation was measured in a liquid scintillation counter.
Patients fulfilled the American Rheumatism Association criteria for classical or definite RA. Serological and cellular typing for determination of class I and class I1 HLA phenotypes was performed on PBMC of the patients. 2.2 Antigens and mitogens The M . tuberculosis ultrasonicate preparation was kindly donated by Paul Klatser from the Royal Tropical Institute and generated as follows. M. tuberculosis strain H 37 RA was ultrasonicated in 0.1 M PBS, pH 7.2, on ice for 16 min at 90 W and 50 duty cycles. The secreted 30/31-kDa proteins corresponding to the antigen 85 complex of BCG were purified from the culture filtrate of M. tuberculosis as previously described [9]. PHA was obtained from the Wellcome Foundation (Beckenham, GB).
2.5 Gel electrophoresis and immunoblotting
The M . tuberculosis ultrasonicate preparation (0.8 mg, 2.4 mg or 8 mg) was applied in a comb of 12 cm length on a 1.55 mm thick 11%SDS-PAGE gel and run for 10 cm under reducing conditions. Samples were obtained by dissolving the proteins in sample buffer and heating them for 5 min at 100°C. Sample buffer consisted of 62.5 mM Tris-HC1 pH 6.8, 2% SDS, 10% glycerol, 0.002% bromophenolblue and 5% 2-ME. After electrophoresis Ag were electrotrans2.3 Generation of M. tuberculosis-reactiveT cell clones ferred to nitrocellulose paper with a 0.2 pm pore size paper T cell clones were isolated from T cell lines that had been (Schleicher & Schuell, Dassel, FRG). Marker proteins generated by stimulation of SFMC of RA patients with 10 were also run on the gel and blotted to a particular part of pglml M. tuberculosis for 5 days and expansion with 10% the nitrocellulose filter which was stained with amido black Lymphocult (Biotest, Dreiech, FRG) for another 7 days. for molecular mass determinations. The nitrocellulose filter Clones from patients W, K and L were generated as follows. was cut into 22 25 x 4 m m strips and each strip was Cells were cultured at a concentration of 0.3 celVwel1 in converted into a microparticulate form as described [lo]; 96-round-bottom-well plates in 100 pl of culture medium the material was solubilized and sterilized in 500 pI DMSO consisting of IMDM with 10% type AB normal human for 1 h at room temperature. Precipitation was performed serum supplemented with 0.1 pg PHA, 10% Lymphocult, 2 by adding 500 pI 0.05 M sodium carbonatelhydrogen carx 10“ allogeneic 3000 rad-irradiated allogeneic PBMC from bonate buffer, pH 9.6 while shaking continuously. Each at least three different individuals and 2 x 103 allogeneic suspension was washed three times with RPMI, resusirradiated (5000 rad) EBV-transformed B cells (EBVB). pended in 1 ml RPMI and kept at -20°C. In a first After 3 days 100 p1 of culture medium supplemented with screening assay clones were tested with immunoblot fractions from the gel loaded with 2.4 mg M . tuberculosis. Most 10% Lymphocult was added per well and after another 4-7 days, cells from positive wells were transferred to macro- clones responded to one or a few adjacent fractions of the wells, restimulated with culture medium containing per ml “2.4 mg panel“. However, a few clones did not react with 1 pg PHA, 10% Lymphocult, lohirradiated PBMC and 105 any fraction of this panel; these were found to be positive irradiated EBVB and expanded in culture medium plus with fractions of the “10 mg panel“. Other clones re10% Lymphocult. Only the Tcell clones of patient H were sponded to all fractions of the “2.4 mg panel“, probably generated by means of an Ag-specific method. Instead of because of some overspill during electrophoresis Ag was PHA and allogeneic feeder cells a M . tuberculosis ultrason- present in all fractions.These clones showed a nice clear-cut icate suspension and autologous peripheral blood cells as a reaction pattern with the fractions of the “0.8 mg source of APC were used in this cloning procedure. panel”.
Table 1. M. fuberculosis-reactive T cell clones from SF of RA patientsa)
Patients
Cloning
M. tuberculosis-reactive Tcell clones
method
DR(Dw) typing
W DRl,DR4(Dw 14) L DRl,DR4(Dwl4) K DRl .DR4(Dw4) H DR4(Dw4.Dw14)
Stimulus
number
restriction
PHA PHA PHA
n = 6 n = 5 n = 14
DR1 (2) or Dw14 (4) DRl (1) or Dw14 (4) DRI (4) or Dw4 (10)
Antigen specific
I1 =
5
Dw4 (1) Dw4 and Dw14 (2) Dw14 (2)
4 M. tuberculosis-reactive Tcell clones were generated
from M. tuberculosis specific SF Tcell lines by LD either in the presence of PHA and allogeneic feeder cells or M. tuberculosis and autologous APC.
Eur. J. Imrnunol. 1991. 21: 1297-1302
M. tuberculosis-reactive synovial fluid T cell clones
1299
SDS-PAGE immunoblot particle suspensions. In this way the molecular mass of the protein(s) recognized by each clone could be assigned to a known range. Of the original 30 clones only 15 were tested. The other 15 either lost their specificity during restimulation cycles or were lost due to infection. As demonstrated in Fig. 1 for clone H54 the majority of clones responded specifically to one immunoblotfraction with low background reactivity to the other fractions, whereas some clones were positive with more than one adjacent immunoblotfraction.
3 Results 3.1 Isolation and restriction analysis of M.tuberculosis-reactive T cell clones
As shown in Table 1, M. tuberculosis-reactive Tcell clones were isolated from M.tuberculosis induced SF Tcell lines of four RA patients, whose SFMC showed a strong proliferative response in the presence of a M. tuberculosis ultrasonicate preparation (data not shown). T cells were cultured under LD conditions either in an Ag-dependent (patient H) or in a mitogen-dependent manner (patients W,K,L). A summary of the immunoblot analysis of the clones is Contrary to expectation, all clones isolated fromTcell lines shown inTable 3.Within the small series of clon,es from each of the patients W,K, and L were M. tuberculosis reactive, patient already many different M. tuberculosis Ag were whereas only 50% of the clones obtained from patient H recognized. Five clones reacted with the immunoblot responded to M. tuberculosis. All four individuals were fraction 12 containing proteins in the molecular mass range D(R) heterozygous and three out of four were DR lDR 4 . of 31-34-kDa. Because these clones recognized proteins in Most individual clones were restricted either via one or the this molecular mass range, we tried to distinguish them by other autologous HLA-DR determinant, but both types of testing their response against the major secreted antigen 85 clones were obtained from each individual. Examples of all complex of M. tuberculosis. This complex is seen as a different proliferationhestriction patterns of the clones that prominent doublet of 30 and 31-kDa on Western blots and we have observed are given in Table2. None of the consists of three closely related components 85A (identical DR1-restricted clones was restricted via DR4 and vice to P32), 85B (identical to a antigen) and 85C with extensive versa. The DR4 restriction was further analyzed by using Dw4+ and Dw14+ APC. Most clones proliferated to M. tuberculosis only when the Ag was presented by accessory cells carrying donor-matched HLA-Dw(4 or 14) subtype determinants (KC8). Some clones also slightly responded when the Ag was presented by DR4-matched but Dw(4 or 14)mismatched determinants (KB8 and W13). Two clones 81 from individual H (43 and 54) responded to the same extent CI whether Dw4+ or Dw14+ APC were used. Like most clones, the T cell lines from which the clones originated showed a much higher reactivity using APC with donormatched DR4 subtypes (data not shown). 21 0
3.2 T cell clone recognition of M. tuberculosis immunoblotfractions
IIlrnmL .mlr Imlr. C
1 2 3 4
t
94kDa
To investigate whether SF Tcells from RA patients display a preferential recognition of certain proteins and whether a relationship with DR1 and/or DR4 restriction can be found, the in vitro proliferative response of the Tcell clones was tested in the presence of matched APC and M. tuberculosis
I . I
5 6 7 8 9 10111213141518171819202122
t
67
t
43
t
30
t
20.1
t
14.4
Figure I. Example of the immunoblot fraction recognition pattern of a M . tuberculosis-reactiveTcellclone.Tcell clone cells (104) and 5 x 104 irradiated PBMC were cultured for 3 days in the presence of 20 pl of an immunoblot particle suspension. C = control fraction containing unloaded nitrocellulose.
Table 2. Restriction patterns of M . tuberculosis-reactiveT cell clonesa)
Clone
M. tubemulosir
APC DR1
KB8 KC8
+-
-
138 355 93 122 53 40613 453
+
nt
+ -
W13
+
w4
+
H54
230
-
-
DR1 Dw4 Dw4 Dw14 427 52037 61183 1193 258 217 117 190 267 14677 19150 352 122 97 140 217 172 763 30627 52 145 127 143 57750 nt nt 135 220 62 nt
nt
383
36153
1w
Dw14
3465
197 90 68
58807 140 467 75 nt
a) Tcells (104) and 5 x 104 DW4+ or Dw14+ irradiated (2000 rad) PBMC as APC were cultured for 3 days in the presence of 10 pg/ml M . tuberculosis.
1300
Eur. J. Immunol. 1991. 21: 1297-1302
P. C. M. Res, D. L. M. Orsini, J. M.Van Laar et al.
Table 3. Immunoblot analysis of M. tuberculosis-reactive T cell clonesa) Clonc
Restriction determinant
W4 w5 w13
12 21 +22 12
WY
DRl DR 1 Dw14 Dw14 Dwl4
LF7 LF5 LG4
Dw14 Dwl.1 Dw14
6 8 20
H38 H32 H13 HS4
Dwll
5 9 + 10 12 12 15
w1s
H16
12 17-20
Dwll
Dw4IDwlJ Dw4IDw 14 Dw4
KBX KC 1
Immunoblot fraction Corresponding recognized molecular mass (kDa)
14 14
DwJ
Dw4
31-34 14- 16 31-34 31-34 16-23 54-60
44-48 16-17 60-71 37-44 31-34 31-34 25-27 27-29 27-29
overlap in antibody recognition and N-terminal amino acid sequence [9, 11, 121. Another option to discriminate the fraction 12 reactive clones was a possible cross-reactivity with M. leprue. As shown inTable 4, the five clones showed three different Ag reactivity patterns on the basis of restriction via different DR molecules and the presence or absence of a proliferative response to M. leprue or to the M. tuberculosis 85 complex. Combining these data with the immunoblot data we conclude that within this series of 15 clones already 12 different Ag specificities can be distinguished. Thus the overall recognition pattern of M. tuberculosis proteins by these DR1- and DRCrestricted synovial T cell clones is quite heterogeneous.
4 Discussion The data presented in this study do not offer evidence for a preferential recognition of a limited number of M. tuberculosis proteins by SF T cells from RA patients. Within a panel of 15 DR1- and DRCrestricted M. tubercufosisreactive T cell clones isolated from the SF of four RA patients already 12 different Ag specificities were present. In this small panel of clones also no relationship between Ag recognition and DR1 and/or DR4 was found. The chance for a M. tuberculosis protein to contain a T cell epitope(s) that is also present on a joint-specific self Ag is
a) Tcells (104) and 5 X 104 irradiated PBMC were cultured for 3 days in the presence of 20 p1of a given immunoblot particle suspension. Three different panels of immunoblot fractions originating from gels loaded with respectively 0.8,2.4 or 8 mg M.tuberculosbwere available for this analysis. In most cases clones were positive with one or a few fractions of the “2.4 mg” panel, but some additional tests with one of the other panels were necessary to define the fraction(s) recognized (see Sect. 2.5).
probably small, therefore the cross-reactivity concept would predict a dominant recognition of only a few M. tuberculosis proteins by synovial T cells. The observed heterogeneity in Ag reactivity of the T cell clones argues against a selective expansion within the joint of only those types of M. tuberculosis-reactive T cells that have crossreacted with joint components. A heterogeneity in immunoblot fraction recognition was also found for nonlesional M. tubercufosis-reactive T cell clones from the peripheral blood of normal individuals and leprosy patients (A. Janson et al., manuscript in preparation). Only 1 out of 13 DRCrestricted T cell clones investigated in our study responded to the immunoblotfraction containing proteins in the 47-kDa range; an M. tuberculosis immunoblot fraction with proteins in this range was shown to be a major M. tuberculosis fraction recognized by peripheral blood Tcells from DR4+ RA patients and controls [8]. None of our TcR/a/P+ T cell clones responded to the immunoblotfraction containing M. tuberculosis proteins with a molecular mass < 10-kDa; this confirms the observation of Pfeffer et al. who demonstrated that only TcR y/6+ Tcells recognize column-separated M. tuberculosis components with an Mr < 10-kDa [13]. In an earlier study we have shown that responses to E. coli were found with similar frequency and magnitude in SFMC as responses to M. tuberculosis [5]. Usually the same samples
Table 4. Diversity in antigen specificity of Tcell clones responding t o M. tuberculosis immunoblot fraction 12 (31-34-kDa)a) (‘lcmc
W4 W13
w15 H.13 HS1
APC
DRI Dwll Dwll Dwl Dwl
No antigen 51
s2 67 383 137
M. tuberculosis
21 31 1 37 977 5 977 22 537
Antigen M. leprae
38 123 52 11650 4 788
M . tuberculosis 85 complex 63 18370 3648 nt 143
a) Tcells (104) and 5 x 104 irradiated PBMC as APC were grown for 3 days in the presence or absence of 10 &ml of a given Ag preparation.
Eur. J. Immunol. 1991. 21: 1297-1302
were positive for both Ag preparations. Therefore, we considered the possibility that at least part of the responses to M. tuberculosis and E. coli are directed to shared epitopes, perhaps expressed by homologous antigens like heat-shock proteins. However, none of the M.tuberculosisreactive clones presented in this report cross-reacts with E. coli (data not shown).Thus the E. coli reactivity of SFMC, which is just as prominent as the reactivity to M. tuberculosis, is directed to other antigens and thus exerted by different synovial T cells. In the same study we found responses to both M. tuberculosis and E. coli to be also present in a high percentage of pleural exudate mononuclear cell samples from non-RA patients with a chronic pleuritis not due tuberculosis. Enhanced Ag reactivity is therefore presumably a general characteristic of mononuclear cells from chronic inflammation sites. The M.tuberculosis-reactive T cell clones were specifically restricted via one HLA-DR or Dw determinant. Thus the Ag recognized by these T cell clones is probably not a superantigen since T cell responses against superantigens, although dependent on HLA class I1 molecules, are not specifically restricted by products of certain alleles [14, 151. The clones described in this study were restricted via DR1 or DR4. DR1 and especially the DR4 subtypes Dw4, Dw14 and Dw15 are restriction determinants that confer susceptibility for I U [16,17].DRl, Dw14 and Dw15 all share the amino acid (aa) 67-74 stretch on the third hypervariable region of the p chain of the HLA-DR molecule, whereas Dw4 only differs at position 71. These molecules differ strongly at this part of the p chain from other DR types and also from the DR4 subtype DwlO, which are not associated with a higher risk to develop RA. According to the so-called shared epitope hypothesis, aa 67-74 of the third hypervariable region of the p chain of DR1, Dw4, Dw14 and Dw15 molecules are causally related to susceptibility to RA [18]. Being part of the binding site for Ag these aa will be directly involved in TcR-Ag interactions critical for the development of RA, presumably as restriction sites for T cells responding to putative autoantigens. The shared epitope hypothesis most likely implies the autoreactive T cells to be restricted via all D(R) molecules that predispose for the development of RA and/or to recognize a similar peptide [MI. Since our clones, being either DR1 or DR4 restricted and probably recognizing different peptides, do not fulfil either of these criteria, it is not likely that the M. tuberculosis-reactive T cells isolated were directly involved in the response to self Ag. However, we do not necessarily have to conclude from our data that M. tuberculosis-reactive T cells do not play any role in the pathogenesis of RA. At least three explanations compatible with this notion are left. First, the in vitro properties of Tcells from the SF may not reflect the properties of cells from the synovium, for instance the CD4/CD8 ratio is usually much lower in the SF than in the synovium. It could be that those M. tuberculosis-reactive T cells that can cross-react to autoantigens are suppressed within the SF, whereas suppression of the same Tcells within the synovium and/or the peripheral blood for some reason is less effective.This would be analogous to the observation that Tcell clones responding to the myelin basic protein can be isolated from the peripheral blood but not from the cerebrospinal fluid of multiple sclerosis patients
M . tuberculosis-reactive synovial fluid T cell clones
1301
[19] and Haffler, personal communication). If the suppression of the disease causing SF Tcells would be mediated by epitopes on M. tuberculosis proteins different from those recognized by Th cells, then the presentation of small peptides containing only helper epitopes instead of whole bacteria might reveal otherwise obscured T cell specificities. Paradoxically, the 65-kDa heat-shock protein is an obvious candidate for inducing suppression of DR4restricted Tcells, because thus far none of the described 27 Tcell epitopes on the 65-kDa protein has been found to be DR4 restricted. Of course, none of the DRCrestricted T cell clones described in this article responds to the 65-kDa heat-shock protein either (data not shown). A second possibility is that M. tuberculosis-reactive SF T cells are directly stimulated by M. tuberculosis antigens present in the joint. In this case a cross-reactivity concept is not necessary, because the joint destruction will be caused by interleukins secreted by activated M. tuberculosisreactiveTcells. M.tuberculosis proteins diffused away from the site of infection might end up in the joint for instance because they bind specifically to joint components. The M. tuberculosis 85 complex is particulary interesting in this respect since it contains at least two proteins with the capability to bind to fibronectin, an extracellular matrix protein abudantly present in joints. The Tcell clones W13 and W15 described in this study are responsive to the M. tuberculosis 85 complex. Finally, a possible role for M. tuberculosis-reactiveTcells in the pathogenesis of RA may be that these cells do not react to autoantigens themselves, but give other Tcells the opportunity to do so. A cellular immune response induced by an infection with M.tuberculosis, which probably shares many homologous aa stretches with self proteins, may somehow interfere with preexisting networks involved in the control of undesired response to self antigens. This might give Tcells specific for endogenous antigens, in this case joint components, the chance to escape from regulation [20]. Perhaps not the T cells responding to self antigens but the M. tuberculosis-reactive Tcells which interact with the regulatory networks could be restricted by D(R) molecules that confer susceptibillity to RA.The M.tuberculosis-reactiveTcel1population with this capacity does not necessarily have to display a limited Ag recognition.
The authors wish to thank Tom Ottenhoff for helpful advice; Paul Klatser of the Royal Tropical Institute in Amsterdamfor performing the SDS-PAGE gel electrophoresis and subsequent blotting; Annemarie Termijtelenforproviding Tcell clones for typing the Dw type of DR4-positive cells; Ieke Schreuder and colleagues for serological HLA typing, and Irun Cohen and Jelle Thole for stimulating discussions. Received January 10, 1991.
5 References 1 Van Eden,W., Holoshitz, J., Nevo, Z., Frenkel, A., Klajman, A. and Cohen, I. R.,Proc. Nail. Acad. Sci. USA 1985. 82: 5117. 2 Van Eden, W.,Thole, J. E . R.,Van der Zee, R . , Noordzij, A., Van Embden, J. D. A., Hensen, E . J. and Cohen, I. R., Nature 1988. 331: 171.
1302
P. C. M. Res, D. L. M.Orsini, J. M.Van Laar et al.
3 Holoshitz, J., Klajman, A., Drucker, I., Lapidot, Z.,Yaretzky, A., Frenkel, A.,Van Eden,W. and Cohen, I. R., Lancet 1986. ii: 305. 4 Res, P. C. M., Schaar, C. G., Breedveld, F. C.,Van Eden,W.,Van Embden, J. D. A., Cohen, I. R. and DeVries, R. R. P., Lancet 1988. ii: 477. 5 Res, P. C. M., Telgt, D.,Van Laar, J. M., Oudkerk Pool, M., Breedveld, F. C. and De Vries, R. R. P., Lancet 1990. 336: 1406. 6 Gaston, J. S. H., Life, P. F., Bailey, L. C. and Bacon, I? A., J. Immunol. 1989. 143: 2494. 7 Soderstrom, K., Halapi, E., Nilsson, E., Gronberg, A., Van Embden, J. D. A., Klareskog, L. and Kiessling, R., Scand. J. Immunol. 1990. 32: 503. 8 Palacios-Boix, A. A., Estrada, G. I., Colston, M. J. and Panayi, G. S., J. Immunol. 1988. 140: 1844. 9 Abou-Zeid, C., Ratliff, T. L., Wiker, H. G., Harboe, M., Bennedsen, J. and Rook, G. A. W., Infect. Immun. 1988. 56: 3046. 10 Abou-Zeid, C., Filley, E., Steele, J. and Rook, G. A. W., J. Immunol. Methods 1987. 85: 5.
Eur. J. Immunol. 1991. 21: 1297-1302
11 De Wit, L., De la Cuvellerie, A., Ooms, J. and Content, J., Nucleic Acids Rex 1990. 18: 3995. 12 Wiker, H. G., Sletten, K., Nagai, S. and Harboe, M., Infect. Immun. 1990. 58: 272. 13 Pfeffer, K., Schoel, B., Gulle, H., Kaufmann, S. H. E. and Wagner, H., Eur. J. Immunol. 1990. 20: 1175. 14 Fleischer, B., Schrezenmeier, H. and Conradt, I?, Cell. Immunol. 1989. 120: 92. 15 Dellabona, P., Peccoud, J., Kappler, J., Marrack, F!, Benoist, C. and Mathis, D., Cell 1990. 62: 1115. 16 Wordsworth, B. P., Lanchbury, J. S. S., Sakkas, L. I.,Welsh, K. I., Panayi, G. S. and Bell, J. I., Proc. Narl. Acad. Sci. USA 1989. 86: 10049. 17 Watanabe,Y. ,Tokunaga, K., Matsuki, K. ,Takeuchi, F., Matsuta, K., Maeda, H., Omoto, K. and Takeo, J. J., J. Exp. Med. 1989. 169: 2263. 18 Gregersen, P. K., Silver, J. and Winchester, R. J., Arthritis Rheum. 1987. 30: 1205. 19 Ota, K., Matsui, M., Milford, E. L., Mackin, G. A.,Weiner, H. L. and Haffler, D. A., Nature 1990. 346: 183. 20 Cohen, I. R. and Young, D. B., Immunol. Today, in press.
