MYCOBACTERIA AND HUMAN HEAT SHOCK PROTEIN-SPECIFIC CYTOTOXIC T LYMPHOCYTES IN RHEUMATOID SYNOVIAL INFLAMMATION SHU GUANG LI, ALISON J. QUAYLE, YAMIN SHEN, JENS KJELDSEN-KRAGH, FREDRIK OFTUNG, RADHEY S. GUPTA, JACOB B. NATVIG, and OYSTEIN T. F@RRE Objective. To study the cytotoxic capacity of mycobacteria-specific T lymphocyte lines and clones from sites of inflammation in patients with rheumatoid arthritis (RA). We also studied antigen specificity, surface phenotype, expression of T cell receptors (TCR), and HLA restriction. Methods. Autologous macrophages (M+) from the synovial membrane (SM), synovial fluid (SF), or peripheral blood (PB) were used as target cells in cytotoxicity assays. Results. All SM and SF cell lines tested thus far have shown specific lysis of the autologous M+ from SM or PB that had been pulsed with BCG (bacillus CalmetteGuerin), but no cytotoxicity when the targets were -
From the Institute of Immunology and Rheumatology, University of Oslo, the Laboratory of Immunology, Institute for Cancer Research, the Norwegian Radium Hospital, and the Oslo Sanitetsforenings Rheumatism Hospital, Oslo, Norway. Supported by the Norwegian Women’s Public Health Organization, the Grethe Harbitz Legacy. the Leon and Norma Hess Legacy for Rheumatic Research, and by Hafslund-Nycomed, Oslo. Shu Guang Li. MD. PhD: Institute of Immunology and Rheumatology. University of Oslo (current address: Specialty Laboratories Inc.. Santa Monica, CA); Alison J. Quayle, MSc, PhD: Institute of Immunology and Rheumatology, University of Oslo; Yamin Shen, MD: Institute of Immunology and Rheumatology, University of Oslo; Jens Kjeldsen-Kragh. MD: Institute of Immunology and Rheumatology, University of Oslo; Fredrik Oftung, PhD: Laboratory of Immunology, Institute for Cancer Research, the Norwegian Radium Hospital; Radhey S. Gupta, PhD: Professor of Biochemistry, Department of Biochemistry, McMaster University, Hamilton. Ontario, Canada; Jacob B. Natvig, MD, PhD: Professor of Medicine, Institute of Immunology and Rheumatology, University of Oslo; gystein T. Fqirre, MD, PhD: Professor of Medicine, Oslo Sanitetsforenings Rheumatism Hospital. Address reprint requests to Shu Guang Li, MD, PhD, Specialty Laboratories Inc., 221 1 Michigan Avenue, Santa Monica, CA 90404-3900. Submitted for publication May 17, 1991;accepted in revised form November 6, 1991. Arthritis and Rheumatism, Vol. 35, No. 3 (March 1992)
pulsed with irrelevant antigens such as tetanus toxoid and Chlamydia. Both CD4+ and CDS+ cells were shown to be involved in the specific cytolysis. The majority of the cytotoxic T lymphocyte (CTL) lines were TCRaI @+ cells. However, both TCRaIP+ and TCRy/S+ clones (TCR 61+)from one RA patient showed antigenspecific lysis. Antigen-specific recognition by a number of CTL lines and clones generated from SF and SM was restricted by HLA-DR molecules. Two Mycobucterium bovis 65-kd heat shock protein (HSP)-specific TCRcul @+ SF T cell clones also lysed M + that had been pulsed with a recombinant human 65-kd HSP. Conclusion. Joint inflammation and destruction might be partly attributable to a cross-reaction of mycobacteria-induced cytotoxic T cells with self HSP.
Although the cause of rheumatoid arthritis (RA) is not known, activated T lymphocytes in the synovial fluid (SF) and synovial membrane (SM)of RA patients may contribute to the pathogenesis of the disease. Evidence has accumulated which indicates that T cells specific for mycobacterial antigens may play a role in chronic arthritis. Studies of the adjuvant arthritis model (experimentally induced by immunization of Lewis rats with Mycobacterium tuberculosis in oil) indicate that autoreactive T cells triggered by mycobacterial antigens can mediate persistent, chronic arthritis or can provide protection from the development of arthritis (1,2). This disease can be transferred to irradiated naive rats by T cells reactive both with an epitope on the 65-kd heat-shock protein (HSP) of M tuberculosis and with cartilage proteoglycans (3,4). Since HSPs may be produced at sites of inflammation (9,it has been speculated that T cells activated by bacterial antigens might conceivably mediate auto-
HSP-SPECIFIC T CELLS IN RA immune disease. It has been observed that mononuclear cells (MNC) derived from the SF of RA patients reacted more with the purified protein derivative of tuberculin (PPD) (6), with an acetoneprecipitable fraction of M tuberculosis (7), and with a recombinant preparation of the Mycobacteriurn bovis 65-kd HSP (8) or Mycobacterium leprae HSP (9,lO) than did MNC from RA peripheral blood (PB). The HLA-DR4-associated regulation of the immune response to M tuberculosis in RA has been demonstrated (11). However, the 65-kd HSP reactivity was more pronounced in patients with reactive arthritis than in those with RA (8,9), and the enhanced reactivity was not confined to mycobacteria, but was also directed against other bacteria (8,12). Recently, Res et a1 (13) reported that the enhanced reactivity against mycobacterial antigens was not a specific feature of MNC from chronically inflamed joints but might be a general characteristic of chronic inflammation. It has been postulated that when autoantigenreactive cytotoxic T cells encounter their specific antigen on the membrane of autologous macrophages (M+) or other cells, lysis of such targets may occur, (possibly) resulting in tissue damage (14). In humans, CD4+ T cell clones reactive with BCG (bacillus Calmette-Guerin) were reported to display cytotoxic activity toward antigen-presenting cells (APC) from BCG-vaccinated healthy subjects (15). In addition, BCG- and 65-kd HSP-induced cytotoxic T lymphocytes (CTL) directed toward the antigen-pulsed Mc$ have been isolated from healthy individuals (1620) and from leprosy patients (16,18) and tuberculosis patients (20). In studies of mice, accumulating evidence has suggested an important role for CTL in the immune response to mycobacteria and other intracellular parasites (21-23). Thus, the involvement of mycobacteria-specific CTL might be a general phenomenon in chronic inflammation. We present here the results of our studies of antigen-specific cytotoxicity displayed by CTL lines and clones derived from inflamed tissues of RA patients. Their antigen specificity, surface phenotype, expression of T cell receptor (TCR), and HLA restriction are described. We also discuss the possible immunopathologic roles of mycobacteria-activated CTL in the synovial inflammation of RA.
MATERIALS AND METHODS Patients. SM, SF, and PB samples were obtained from patients with RA who were hospitalized at Oslo Sani-
27 1
tetsforenings Rheumatism Hospital. Nine patients (2 males and 7 females) were classified as having RA, according to the 1987 revised criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) (24). The average age of the study population was 56 years. and the average disease duration was 12 years. Antigens. BCG and Mycobacterium kansasii (MK) that had been cultured in synthetic Sauton medium were kindly provided by Dr. H. G. Wiker (Oslo, Norway) and Dr. Sadadu Nagai (Osaka, Japan). The purified recombinant M bovis 65-kd HSP was provided by Dr. Ruurd van der Zee (supported by the United Nations Development Project/ World BanMWorld Health Organization special program for research and training in tropical diseases; the National Institute of Public Health and Environmental Protection, Bilthoven, The Netherlands). Recombinant P1 protein, a human (Hu) 65-kd HSP, was produced by cloning complementary DNA (cDNA) for the mature form of this protein at the Nco I site in the prokaryotic expression vector pKK233-2, as described previously (25.26). The protein preparation was obtained by electroelution from sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The protein and a control preparation from a culture supernatant of Escherichia coli were further purified on nitrocellulose by Dr. M. Sioud (27). Tetanus toxoid (TT) and Chlamydia were kindly provided by the Statens Institutt for Folkehelse (Oslo, Norway). Generation of BCG-stimulated T cell lines from PB, SF, and SM from RA patients. MNC were isolated as previously described from PB, SM, and SF obtained from RA patients (28). Throughout these studies we used RPMI 1640 supplemented with 2 mM glutamine, 100 unitslml penicillin, 100 pdml streptomycin, and 10% human AB serum (complete RPMI). MNC were cultured at lo6 cells/ well for 5 days in 24-well tissue culture plates (Falcon 3047; Becton Dickinson, Mountain View, CA), with 10 pg/rnl BCG, at 37°C in a fully humidified 5% CO, incubator. On day 4, medium containing 20% (volume/volume) interleukin-2 (IL-Zjrich supernatant made from a 48-hour culture of phytohemagglutinin (PHA)-stimulated PBMC from 8 healthy blood donors was added to the cultures. After an additional 4-7 days, the cultures were restimulated with the antigen and irradiated allogenic PB MNC mixture as feeders for the enrichment of the cells. T cell cloning. The BCG-specific TCRa/P+ T cell clones were generated in the following way. SF cells from patient H were stimulated for 7 days with 10 pdml of BCG and then cloned by limiting dilution in 96-well plates, using a mixture of autologous irradiated S F and PB MNC as APC, 20 pg/ml of BCG, and 50 unitslml of recombinant IL-2 (rIL-2). Fresh complete RPMI and rIL-2 were added to the cloning plates after 7 and 12 days. Between days 14 and 20, each clone was transferred to 1 well of a 24-well plate and restimulated with allogeneic PBMC, PHA, and 10% (vlv) IL-2-rich supernatant. Clones were expanded by a twiceweekly addition of complete RPMI and 1L-2 supernatant, and restimulation with allogeneic PBMC and PHA every 12 days (29). None of the clones generated in this way responded to the 65-kd HSP. TCRaIP+ clones specific for 65-kd HSP were generated from patient H according to the same protocol, but with
272
LI ET AL Table 1.
Monoclonal antibodies used
Specificity
Name
IgG isotype
Source
CD3 CD4 CDX CD19 CD45RA CD45RO TCRaIP TCR V,5 family TCR V& family TCR V,8 family TCR V,12 family TCRyI6 HLA class 1 HLA-DR HLA-DQ H LA-DP IgD
IOT3 Anti-Leu-3a Anti-Leu-2a HD237 2H4-RDI UCHLI TCR- I ,V5 (a) ,V6 (a) ,V8 (a) ,V12 (a) TCR-61 B9.12.1 B8.11.2 Anti-Leu-I0 Anti-human HLA-DP TI B96
IgG I IgG I IgG I IgGZb IgG 1 IgG2a IgG I IgG 1 IgG I lgG2b lgG2a IgG I lgG2a lgG2b IgG I IgG I IgG I
Imrnunotech, Paris, France Becton Dickinson. Mountain View, CA Becton Dickinson, Mountain View, CA Ref. 31 Coulter. Hialeah, FL Dakopatt s, Copen hagen, Denmark Becton Dickinson, Mountain View, CA T Cell Sciences, Cambridge, MA T Cell Sciences, Cambridge, MA T Cell Sciences, Cambridge, MA T Cell Sciences, Cambridge, MA T Cell Sciences, Cambridge, MA Ref. 32 Ref. 32 Becton Dickinson, Mountain View, CA Becton Dickinson, Mountain View, CA Ref. 33
15 Fglml65-kd HSP for the 7-day stimulation of S F cells, and then 20 pg/ml of BCG and 2 pglml65-kd HSP in the cloning. The 65-kd HSP-specific TCRy/G+ clones were generated from the synovial fluid of patient I. First, a 65-kd-specific T cell line was established. The cell line was depleted of CD4+ cells and TCRaIP+ cells by use of magnetic beads (30), and the remaining cell population was then cloned by limiting dilution (Kjeldsen-Kragh J et al: unpublished observations). Depletion of lymphocyte subsets using magnetic beads. Commercial magnetic beads (Dynabeads; Dynal, Oslo, Norway) precoated with anti-CD4 and anti-CD8 monoclonal antibodies (MAb) were used for depletion experiments according to the method of Lea et al (30). Monoclonal antibodies. The MAb used in fluorescence-activated cell sorter (FACS) analysis and for inhibition of antigen-specific cytotoxicity are listed in Table 1. Cytotoxicity assay. The use of adherent M d as target cells has been described by Ottenhoff et al (16). Briefly, MNC from PB andlor from SM or SF (10' cells/well) were plated in 96-well U-bottom tissue culture plates (Nunc, Raskilde, Denmark). Approximately 10% of the cells adhered. This number was used for calculating effector:target (E:T) cell ratios. Cells were cultured for 6-14 days, and during the last 18 hours, were pulsed with various antigens (see Results) and labeled with 2 pCi/well of "Cr. Cells were washed 3 times with preheated medium, and the effector cells were added to the target cells (150 pllwell) in duplicate or in triplicate. One triplicate of target cells in medium without effector cells was used to determine the level of spontaneous release. After 16 hours, the content of each well was transferred to a detachable counting tube, and 100 pl of 1% Triton X-100 (no. T-6878; Sigma, St. Louis, MO) was added to the original wells to lyse the remaining adherent cells. After 3 hours, the total volume of Triton was transferred to similar tubes, and the samples were counted in a gamma counter (model 1275 minigamma; LKB, Stockholm, Sweden). The percentage of specific "Cr release for each well was calculated as follows: 3' 6 specific lysis = [test counts per minute/(test cpm + cpm after Triton X-100treatment of the same well)] x 100% - % spontaneous release. The percent-
age of spontaneous release was calculated as follows: cpm in spontaneous release well/(cpm in spontaneous release well + cpm after Triton X-100treatment of the same well) x 100%. Phenotypic analysis of effector cells. T cell lines and clones were stained by using a standard direct andlor indirect immunofluorescence technique, and were analyzed by a FACScan flow cytometer (Becton Dickinson) (34).
RESULTS Cytotoxicity of BCG-specific T cells from SF and SM. As shown in Figure 1, BCG-stimulated lymphocytes from the SM (Figure 1A) and the SF (Figure 1B) of RA patients lysed BCG-pulsed autologous M 4 in a dose-dependent manner. The antigen-dependent lysis was specific; it was not accompanied by lysis of nonpulsed target cells or of target cells pulsed with irrelevant antigens (TT or Chlamydia). The background killing of nonpulsed M 4 was between 2% and 35%, depending on the individual; this level is consistent with those in other published studies (16-18,20). A panel of antigens was used in this cytotoxicity assay to test for antigen specificity. A BCG-stimulated SF T cell clone from patient H, USSF2B9 (Figure IB), showed cross-reactivity between BCG and MK, but not with unrelated antigens (TT or Chlamydia) or with a recombinant M bovis 65-kd HSP. To detect false lytic activity, we included both positive and negative controls in our assay system. The positive lysis control was a human HLA-DR4/ Dw4-specific CTL cell line generated from a random donor against an HLA-DR4/Dw4 positive mouse transfectant (Shen Y et al: unpublished observation) lysed both BCG-pulsed and unpulsed targets (HLA-
273
HSP-SPECIFIC T CELLS IN RA
*O/
40
A
t 0
7 7
t
10
20
30
c
-
2ol'O
0
5
10
loor D
t
30t
40
I
Figure ID shows such an example by using a T cell clone from patient H. This was used as the negative lysis control. To confirm the above observations, more RA patients were tested. Figure 2 shows that SM and SF T lymphocyte lines and clones from 9 RA patients, which were raised after stimulation with BCG for short-term (7 days to 1 month) or long-term (>12 months) culture, significantly lysed BCG-pulsed M+ from SM, SF, or PB as compared with antigenunpulsed and TT-pulsed targets. Furthermore, all BCG-induced T cell lines tested from patients A 4 were cytotoxic neither for K562 cells (the conventional natural killer [NK] target) nor for Daudi cells (resistant to NK cells but sensitive to activated killer [AK] activity target) (data not shown). The results suggest that the mycobacteriainduced cytotoxicity is not an NWAK-like activity. Phenotype of specific CTL originating from SM and SF. The phenotyping data from 7 BCG-stimulated SM CTL lines from 7 RA patients are shown in Table 2. Two well-defined 65-kd HSP-specific TCRaIP+ SF T cell clones derived from patient H and 1 TCRyIG+ 60
0
10
20
30
0
1.5
15
E/T ratio Figure 1. Antigen-specific cytotoxicity of bacillus Calmette-Guerin (BCGtstimulated T effector cells from the synovial membrane (SM) and the synovial fluid (SF) of rheumatoid arthritis (RA) patients, against antigen-pulsed autologous macrophages (MI$). A, BCGstimulated SM T cell line from patient A, cytotoxic against BCGpulsed (O),but not tetanus toxoid ('ITtpulsed (A) or unpulsed (O), autologous M 4 from SM. B, BCG-stimulated S F T cell clone (USSF2B9) from patient H, cytotoxic to autologous MI$ from peripheral blood (PB) pulsed with BCG (0)and Mycobacterium kansasii (0),but not TT (A),Chlamydia (A), recombinant Mycobacterium bovis 65-kd heat-shock protein (HSP) (W), or unpulsed (0)cells. C, A human cytotoxic T lymphocyte (CTL) line against mouse HLA-DR4/Dw4 positive transfectant generated from a random donor (HLA-DR7, 10; DQI, 2) lysed PB M 4 from patient H (HLA-DR4/Dw4, DR5), both when pulsed with BCG (0)and when unpulsed (without antigen) (0).D, Antigen-specific cytotoxicity of M bovis 65-kd HSP-stimulated SF T cell clone (USSF2C6 from patient H; HLA-DR4,5/Dw4) against BCG-pulsed autologous MI$ (O),but not unpulsed autologous MI$ (0) from . PB. It did not lyse BCG-pulsed (Vand +) or unpulsed (V and 0)allogeneic MI$from the PBofdonor I (VandV)(HLA-DRI, 1; DQ1, w5,6)ordonor2(+and 0 )(HLA-DRS, w6, DRw52; DQI. 3, w6, 7). Em = effectorltarget.
DR4Dw4 and DR5) equally (Figure 1C). In contrast, BCG-stimulated SM as well as SF effector cells did not lyse BCG-pulsed, HLA-mismatched allogeneic M+.
-a?
>.
.0 .-
50
40
c
x
0 0
30
c
20
0
ia a
A
B
C
D
E
F
G
H
I
RA patient
Figure 2. BCG-activated SM T cell lines (A-G) and SF T cell clones (H-I) of 9 RA patients lyse BCG-pulsed, but not TT-pulsed or unpulsed autologous M+ from PB (A, B, E, H, and 1) or SM (C, D, F, and G). Values are the percentage of mean 5'Cr release in triplicate cultures; the SEM 5'Cr release did not exceed 15%. The E m ratio was 7:l for patient 1, 15:l for patient G, 20:l for patients C-E, 301 for patient A and H, 40:l for patient F, and 50:1 for patient B. Pulsing of MI$ did not affect viability of the M 4 in the absence of effector cells. Thus, the spontaneous "Cr release of antigen-unpulsed M 4 from PB was 24.6% (n = 5 ; SEM 2.2%) and that of M 4 from SM was 28% (n = 4; SEM 2.1%) after 18 hours of incubation, release of BCG-pulsed MI$ from PB was 24.6% (SEM 2.6%) and that of M 4 from SM was 28.3% (SEM 3.3%), release of 'IT-pulsed M4from PB was 27.2% (SEM 2.5%) and that of M4frorn SM was 28.8% (SEM 3.3%). Ag = antigen (see Figure 1 for other definitions).
LI ET AL
274
Table 2. Phenotypes of BCG-stimulated T cell lines (A-G) and 65-kd HSP-induced clones (H and 1) derived from 9 RA patients* % labeling of TCR types
9% labeling of CD MAb
Patient
3
4
8
45RO
alp
V5
VB6
V$
3 3 3 5 3 5 2
83 95 84
95 76 92 93
2
94
6 I 7 I 6 2 3
11
94 91 89
9 2 3 2 2 5 9
45RA
% labeling of
V012
Y/6
HLA-DR
15
1
6 7 12
21
~~
A B C D E F G H
98 96 90 91 91 89 84
59 63 62
Clone I Clone 2
97 99
90 96
2
From the Institute of Immunology and Rheumatology, University of Oslo, the Laboratory of Immunology, Institute for Cancer Research, the Norwegian Radium Hospital, and the Oslo Sanitetsforenings Rheumatism Hospital, Oslo, Norway. Supported by the Norwegian Women’s Public Health Organization, the Grethe Harbitz Legacy. the Leon and Norma Hess Legacy for Rheumatic Research, and by Hafslund-Nycomed, Oslo. Shu Guang Li. MD. PhD: Institute of Immunology and Rheumatology. University of Oslo (current address: Specialty Laboratories Inc.. Santa Monica, CA); Alison J. Quayle, MSc, PhD: Institute of Immunology and Rheumatology, University of Oslo; Yamin Shen, MD: Institute of Immunology and Rheumatology, University of Oslo; Jens Kjeldsen-Kragh. MD: Institute of Immunology and Rheumatology, University of Oslo; Fredrik Oftung, PhD: Laboratory of Immunology, Institute for Cancer Research, the Norwegian Radium Hospital; Radhey S. Gupta, PhD: Professor of Biochemistry, Department of Biochemistry, McMaster University, Hamilton. Ontario, Canada; Jacob B. Natvig, MD, PhD: Professor of Medicine, Institute of Immunology and Rheumatology, University of Oslo; gystein T. Fqirre, MD, PhD: Professor of Medicine, Oslo Sanitetsforenings Rheumatism Hospital. Address reprint requests to Shu Guang Li, MD, PhD, Specialty Laboratories Inc., 221 1 Michigan Avenue, Santa Monica, CA 90404-3900. Submitted for publication May 17, 1991;accepted in revised form November 6, 1991. Arthritis and Rheumatism, Vol. 35, No. 3 (March 1992)
pulsed with irrelevant antigens such as tetanus toxoid and Chlamydia. Both CD4+ and CDS+ cells were shown to be involved in the specific cytolysis. The majority of the cytotoxic T lymphocyte (CTL) lines were TCRaI @+ cells. However, both TCRaIP+ and TCRy/S+ clones (TCR 61+)from one RA patient showed antigenspecific lysis. Antigen-specific recognition by a number of CTL lines and clones generated from SF and SM was restricted by HLA-DR molecules. Two Mycobucterium bovis 65-kd heat shock protein (HSP)-specific TCRcul @+ SF T cell clones also lysed M + that had been pulsed with a recombinant human 65-kd HSP. Conclusion. Joint inflammation and destruction might be partly attributable to a cross-reaction of mycobacteria-induced cytotoxic T cells with self HSP.
Although the cause of rheumatoid arthritis (RA) is not known, activated T lymphocytes in the synovial fluid (SF) and synovial membrane (SM)of RA patients may contribute to the pathogenesis of the disease. Evidence has accumulated which indicates that T cells specific for mycobacterial antigens may play a role in chronic arthritis. Studies of the adjuvant arthritis model (experimentally induced by immunization of Lewis rats with Mycobacterium tuberculosis in oil) indicate that autoreactive T cells triggered by mycobacterial antigens can mediate persistent, chronic arthritis or can provide protection from the development of arthritis (1,2). This disease can be transferred to irradiated naive rats by T cells reactive both with an epitope on the 65-kd heat-shock protein (HSP) of M tuberculosis and with cartilage proteoglycans (3,4). Since HSPs may be produced at sites of inflammation (9,it has been speculated that T cells activated by bacterial antigens might conceivably mediate auto-
HSP-SPECIFIC T CELLS IN RA immune disease. It has been observed that mononuclear cells (MNC) derived from the SF of RA patients reacted more with the purified protein derivative of tuberculin (PPD) (6), with an acetoneprecipitable fraction of M tuberculosis (7), and with a recombinant preparation of the Mycobacteriurn bovis 65-kd HSP (8) or Mycobacterium leprae HSP (9,lO) than did MNC from RA peripheral blood (PB). The HLA-DR4-associated regulation of the immune response to M tuberculosis in RA has been demonstrated (11). However, the 65-kd HSP reactivity was more pronounced in patients with reactive arthritis than in those with RA (8,9), and the enhanced reactivity was not confined to mycobacteria, but was also directed against other bacteria (8,12). Recently, Res et a1 (13) reported that the enhanced reactivity against mycobacterial antigens was not a specific feature of MNC from chronically inflamed joints but might be a general characteristic of chronic inflammation. It has been postulated that when autoantigenreactive cytotoxic T cells encounter their specific antigen on the membrane of autologous macrophages (M+) or other cells, lysis of such targets may occur, (possibly) resulting in tissue damage (14). In humans, CD4+ T cell clones reactive with BCG (bacillus Calmette-Guerin) were reported to display cytotoxic activity toward antigen-presenting cells (APC) from BCG-vaccinated healthy subjects (15). In addition, BCG- and 65-kd HSP-induced cytotoxic T lymphocytes (CTL) directed toward the antigen-pulsed Mc$ have been isolated from healthy individuals (1620) and from leprosy patients (16,18) and tuberculosis patients (20). In studies of mice, accumulating evidence has suggested an important role for CTL in the immune response to mycobacteria and other intracellular parasites (21-23). Thus, the involvement of mycobacteria-specific CTL might be a general phenomenon in chronic inflammation. We present here the results of our studies of antigen-specific cytotoxicity displayed by CTL lines and clones derived from inflamed tissues of RA patients. Their antigen specificity, surface phenotype, expression of T cell receptor (TCR), and HLA restriction are described. We also discuss the possible immunopathologic roles of mycobacteria-activated CTL in the synovial inflammation of RA.
MATERIALS AND METHODS Patients. SM, SF, and PB samples were obtained from patients with RA who were hospitalized at Oslo Sani-
27 1
tetsforenings Rheumatism Hospital. Nine patients (2 males and 7 females) were classified as having RA, according to the 1987 revised criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) (24). The average age of the study population was 56 years. and the average disease duration was 12 years. Antigens. BCG and Mycobacterium kansasii (MK) that had been cultured in synthetic Sauton medium were kindly provided by Dr. H. G. Wiker (Oslo, Norway) and Dr. Sadadu Nagai (Osaka, Japan). The purified recombinant M bovis 65-kd HSP was provided by Dr. Ruurd van der Zee (supported by the United Nations Development Project/ World BanMWorld Health Organization special program for research and training in tropical diseases; the National Institute of Public Health and Environmental Protection, Bilthoven, The Netherlands). Recombinant P1 protein, a human (Hu) 65-kd HSP, was produced by cloning complementary DNA (cDNA) for the mature form of this protein at the Nco I site in the prokaryotic expression vector pKK233-2, as described previously (25.26). The protein preparation was obtained by electroelution from sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The protein and a control preparation from a culture supernatant of Escherichia coli were further purified on nitrocellulose by Dr. M. Sioud (27). Tetanus toxoid (TT) and Chlamydia were kindly provided by the Statens Institutt for Folkehelse (Oslo, Norway). Generation of BCG-stimulated T cell lines from PB, SF, and SM from RA patients. MNC were isolated as previously described from PB, SM, and SF obtained from RA patients (28). Throughout these studies we used RPMI 1640 supplemented with 2 mM glutamine, 100 unitslml penicillin, 100 pdml streptomycin, and 10% human AB serum (complete RPMI). MNC were cultured at lo6 cells/ well for 5 days in 24-well tissue culture plates (Falcon 3047; Becton Dickinson, Mountain View, CA), with 10 pg/rnl BCG, at 37°C in a fully humidified 5% CO, incubator. On day 4, medium containing 20% (volume/volume) interleukin-2 (IL-Zjrich supernatant made from a 48-hour culture of phytohemagglutinin (PHA)-stimulated PBMC from 8 healthy blood donors was added to the cultures. After an additional 4-7 days, the cultures were restimulated with the antigen and irradiated allogenic PB MNC mixture as feeders for the enrichment of the cells. T cell cloning. The BCG-specific TCRa/P+ T cell clones were generated in the following way. SF cells from patient H were stimulated for 7 days with 10 pdml of BCG and then cloned by limiting dilution in 96-well plates, using a mixture of autologous irradiated S F and PB MNC as APC, 20 pg/ml of BCG, and 50 unitslml of recombinant IL-2 (rIL-2). Fresh complete RPMI and rIL-2 were added to the cloning plates after 7 and 12 days. Between days 14 and 20, each clone was transferred to 1 well of a 24-well plate and restimulated with allogeneic PBMC, PHA, and 10% (vlv) IL-2-rich supernatant. Clones were expanded by a twiceweekly addition of complete RPMI and 1L-2 supernatant, and restimulation with allogeneic PBMC and PHA every 12 days (29). None of the clones generated in this way responded to the 65-kd HSP. TCRaIP+ clones specific for 65-kd HSP were generated from patient H according to the same protocol, but with
272
LI ET AL Table 1.
Monoclonal antibodies used
Specificity
Name
IgG isotype
Source
CD3 CD4 CDX CD19 CD45RA CD45RO TCRaIP TCR V,5 family TCR V& family TCR V,8 family TCR V,12 family TCRyI6 HLA class 1 HLA-DR HLA-DQ H LA-DP IgD
IOT3 Anti-Leu-3a Anti-Leu-2a HD237 2H4-RDI UCHLI TCR- I ,V5 (a) ,V6 (a) ,V8 (a) ,V12 (a) TCR-61 B9.12.1 B8.11.2 Anti-Leu-I0 Anti-human HLA-DP TI B96
IgG I IgG I IgG I IgGZb IgG 1 IgG2a IgG I IgG 1 IgG I lgG2b lgG2a IgG I lgG2a lgG2b IgG I IgG I IgG I
Imrnunotech, Paris, France Becton Dickinson. Mountain View, CA Becton Dickinson, Mountain View, CA Ref. 31 Coulter. Hialeah, FL Dakopatt s, Copen hagen, Denmark Becton Dickinson, Mountain View, CA T Cell Sciences, Cambridge, MA T Cell Sciences, Cambridge, MA T Cell Sciences, Cambridge, MA T Cell Sciences, Cambridge, MA T Cell Sciences, Cambridge, MA Ref. 32 Ref. 32 Becton Dickinson, Mountain View, CA Becton Dickinson, Mountain View, CA Ref. 33
15 Fglml65-kd HSP for the 7-day stimulation of S F cells, and then 20 pg/ml of BCG and 2 pglml65-kd HSP in the cloning. The 65-kd HSP-specific TCRy/G+ clones were generated from the synovial fluid of patient I. First, a 65-kd-specific T cell line was established. The cell line was depleted of CD4+ cells and TCRaIP+ cells by use of magnetic beads (30), and the remaining cell population was then cloned by limiting dilution (Kjeldsen-Kragh J et al: unpublished observations). Depletion of lymphocyte subsets using magnetic beads. Commercial magnetic beads (Dynabeads; Dynal, Oslo, Norway) precoated with anti-CD4 and anti-CD8 monoclonal antibodies (MAb) were used for depletion experiments according to the method of Lea et al (30). Monoclonal antibodies. The MAb used in fluorescence-activated cell sorter (FACS) analysis and for inhibition of antigen-specific cytotoxicity are listed in Table 1. Cytotoxicity assay. The use of adherent M d as target cells has been described by Ottenhoff et al (16). Briefly, MNC from PB andlor from SM or SF (10' cells/well) were plated in 96-well U-bottom tissue culture plates (Nunc, Raskilde, Denmark). Approximately 10% of the cells adhered. This number was used for calculating effector:target (E:T) cell ratios. Cells were cultured for 6-14 days, and during the last 18 hours, were pulsed with various antigens (see Results) and labeled with 2 pCi/well of "Cr. Cells were washed 3 times with preheated medium, and the effector cells were added to the target cells (150 pllwell) in duplicate or in triplicate. One triplicate of target cells in medium without effector cells was used to determine the level of spontaneous release. After 16 hours, the content of each well was transferred to a detachable counting tube, and 100 pl of 1% Triton X-100 (no. T-6878; Sigma, St. Louis, MO) was added to the original wells to lyse the remaining adherent cells. After 3 hours, the total volume of Triton was transferred to similar tubes, and the samples were counted in a gamma counter (model 1275 minigamma; LKB, Stockholm, Sweden). The percentage of specific "Cr release for each well was calculated as follows: 3' 6 specific lysis = [test counts per minute/(test cpm + cpm after Triton X-100treatment of the same well)] x 100% - % spontaneous release. The percent-
age of spontaneous release was calculated as follows: cpm in spontaneous release well/(cpm in spontaneous release well + cpm after Triton X-100treatment of the same well) x 100%. Phenotypic analysis of effector cells. T cell lines and clones were stained by using a standard direct andlor indirect immunofluorescence technique, and were analyzed by a FACScan flow cytometer (Becton Dickinson) (34).
RESULTS Cytotoxicity of BCG-specific T cells from SF and SM. As shown in Figure 1, BCG-stimulated lymphocytes from the SM (Figure 1A) and the SF (Figure 1B) of RA patients lysed BCG-pulsed autologous M 4 in a dose-dependent manner. The antigen-dependent lysis was specific; it was not accompanied by lysis of nonpulsed target cells or of target cells pulsed with irrelevant antigens (TT or Chlamydia). The background killing of nonpulsed M 4 was between 2% and 35%, depending on the individual; this level is consistent with those in other published studies (16-18,20). A panel of antigens was used in this cytotoxicity assay to test for antigen specificity. A BCG-stimulated SF T cell clone from patient H, USSF2B9 (Figure IB), showed cross-reactivity between BCG and MK, but not with unrelated antigens (TT or Chlamydia) or with a recombinant M bovis 65-kd HSP. To detect false lytic activity, we included both positive and negative controls in our assay system. The positive lysis control was a human HLA-DR4/ Dw4-specific CTL cell line generated from a random donor against an HLA-DR4/Dw4 positive mouse transfectant (Shen Y et al: unpublished observation) lysed both BCG-pulsed and unpulsed targets (HLA-
273
HSP-SPECIFIC T CELLS IN RA
*O/
40
A
t 0
7 7
t
10
20
30
c
-
2ol'O
0
5
10
loor D
t
30t
40
I
Figure ID shows such an example by using a T cell clone from patient H. This was used as the negative lysis control. To confirm the above observations, more RA patients were tested. Figure 2 shows that SM and SF T lymphocyte lines and clones from 9 RA patients, which were raised after stimulation with BCG for short-term (7 days to 1 month) or long-term (>12 months) culture, significantly lysed BCG-pulsed M+ from SM, SF, or PB as compared with antigenunpulsed and TT-pulsed targets. Furthermore, all BCG-induced T cell lines tested from patients A 4 were cytotoxic neither for K562 cells (the conventional natural killer [NK] target) nor for Daudi cells (resistant to NK cells but sensitive to activated killer [AK] activity target) (data not shown). The results suggest that the mycobacteriainduced cytotoxicity is not an NWAK-like activity. Phenotype of specific CTL originating from SM and SF. The phenotyping data from 7 BCG-stimulated SM CTL lines from 7 RA patients are shown in Table 2. Two well-defined 65-kd HSP-specific TCRaIP+ SF T cell clones derived from patient H and 1 TCRyIG+ 60
0
10
20
30
0
1.5
15
E/T ratio Figure 1. Antigen-specific cytotoxicity of bacillus Calmette-Guerin (BCGtstimulated T effector cells from the synovial membrane (SM) and the synovial fluid (SF) of rheumatoid arthritis (RA) patients, against antigen-pulsed autologous macrophages (MI$). A, BCGstimulated SM T cell line from patient A, cytotoxic against BCGpulsed (O),but not tetanus toxoid ('ITtpulsed (A) or unpulsed (O), autologous M 4 from SM. B, BCG-stimulated S F T cell clone (USSF2B9) from patient H, cytotoxic to autologous MI$ from peripheral blood (PB) pulsed with BCG (0)and Mycobacterium kansasii (0),but not TT (A),Chlamydia (A), recombinant Mycobacterium bovis 65-kd heat-shock protein (HSP) (W), or unpulsed (0)cells. C, A human cytotoxic T lymphocyte (CTL) line against mouse HLA-DR4/Dw4 positive transfectant generated from a random donor (HLA-DR7, 10; DQI, 2) lysed PB M 4 from patient H (HLA-DR4/Dw4, DR5), both when pulsed with BCG (0)and when unpulsed (without antigen) (0).D, Antigen-specific cytotoxicity of M bovis 65-kd HSP-stimulated SF T cell clone (USSF2C6 from patient H; HLA-DR4,5/Dw4) against BCG-pulsed autologous MI$ (O),but not unpulsed autologous MI$ (0) from . PB. It did not lyse BCG-pulsed (Vand +) or unpulsed (V and 0)allogeneic MI$from the PBofdonor I (VandV)(HLA-DRI, 1; DQ1, w5,6)ordonor2(+and 0 )(HLA-DRS, w6, DRw52; DQI. 3, w6, 7). Em = effectorltarget.
DR4Dw4 and DR5) equally (Figure 1C). In contrast, BCG-stimulated SM as well as SF effector cells did not lyse BCG-pulsed, HLA-mismatched allogeneic M+.
-a?
>.
.0 .-
50
40
c
x
0 0
30
c
20
0
ia a
A
B
C
D
E
F
G
H
I
RA patient
Figure 2. BCG-activated SM T cell lines (A-G) and SF T cell clones (H-I) of 9 RA patients lyse BCG-pulsed, but not TT-pulsed or unpulsed autologous M+ from PB (A, B, E, H, and 1) or SM (C, D, F, and G). Values are the percentage of mean 5'Cr release in triplicate cultures; the SEM 5'Cr release did not exceed 15%. The E m ratio was 7:l for patient 1, 15:l for patient G, 20:l for patients C-E, 301 for patient A and H, 40:l for patient F, and 50:1 for patient B. Pulsing of MI$ did not affect viability of the M 4 in the absence of effector cells. Thus, the spontaneous "Cr release of antigen-unpulsed M 4 from PB was 24.6% (n = 5 ; SEM 2.2%) and that of M 4 from SM was 28% (n = 4; SEM 2.1%) after 18 hours of incubation, release of BCG-pulsed MI$ from PB was 24.6% (SEM 2.6%) and that of M 4 from SM was 28.3% (SEM 3.3%), release of 'IT-pulsed M4from PB was 27.2% (SEM 2.5%) and that of M4frorn SM was 28.8% (SEM 3.3%). Ag = antigen (see Figure 1 for other definitions).
LI ET AL
274
Table 2. Phenotypes of BCG-stimulated T cell lines (A-G) and 65-kd HSP-induced clones (H and 1) derived from 9 RA patients* % labeling of TCR types
9% labeling of CD MAb
Patient
3
4
8
45RO
alp
V5
VB6
V$
3 3 3 5 3 5 2
83 95 84
95 76 92 93
2
94
6 I 7 I 6 2 3
11
94 91 89
9 2 3 2 2 5 9
45RA
% labeling of
V012
Y/6
HLA-DR
15
1
6 7 12
21
~~
A B C D E F G H
98 96 90 91 91 89 84
59 63 62
Clone I Clone 2
97 99
90 96
2
