PANDR-Binding Hsp60 self epitopes induce an interleukin-10mediated immune response in rheumatoid arthritis.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 60, No. 7, July 2009, pp 1966–1976 DOI 10.1002/art.24656 © 2009, American College of Rheumatology Pan–DR-Binding Hsp60 Self Epitopes Induce an Interleukin-10–Mediated Immune Response in Rheumatoid Arthritis Huib de Jong,1 Floris F. P. Lafeber,1 Wilco de Jager,1 Margje H. Haverkamp,1 Wietse Kuis,1 Johannes W. J. Bijlsma,1 Berent J. Prakken,2 and Salvatore Albani3 Objective. Human Hsp60 is expressed in the joints of patients with rheumatoid arthritis (RA) and can elicit a regulatory T cell response in the peripheral blood and synovial fluid. However, Hsp60 can also trigger strong proinflammatory pathways. Thus, to understand the nature of these Hsp60-directed responses in RA, it is necessary to study such responses at the molecular, epitope-specific level. This study was undertaken to characterize the disease specificity and function of pan–DR-binding Hsp60–derived epitopes as possible modulators of autoimmune inflammation in RA. Methods. Lymphocyte proliferation assays (using 3 H-thymidine incorporation and carboxyfluorescein diacetate succinimidyl ester [CFSE] staining) and measurement of cytokine production (using multiplex immunoassay and intracellular staining) were performed after in vitro activation of peripheral blood mono- nuclear cells from patients with RA, compared with healthy controls. Results. A disease (RA)–specific immune recognition, characterized by T cell proliferation as well as increased production of tumor necrosis factor ␣ (TNF␣), interleukin-1␤ (IL-1␤), and IL-10, was found for 3 of the 8 selected peptides in patients with RA as compared with healthy controls (P < 0.05). Intracellular cytokine staining and CFSE labeling showed that CD4ⴙ T cells were the subset primarily responsible for both the T cell proliferation and the cytokine production in RA. Interestingly, the human peptides had a remarkably different phenotype, with a 5–10-fold higher IL-10:TNF␣ ratio, compared with that of the microbial peptides. Conclusion. These results suggest a diseasespecific immune-modulatory role of epitope-specific T cells in the inflammatory processes of RA. Therefore, these pan–DR-binding epitopes could be used as a tool to study the autoreactive T cell response in RA and might be suitable candidates for use in immunotherapy. Dr. Prakken’s work was supported by a Vidi innovation grant from The Netherlands Organization for Scientific Research, and by the Dutch Arthritis Foundation. 1 Huib de Jong, MD, Floris F. P. Lafeber, PhD, Wilco de Jager, PhD, Margje H. Haverkamp, MD, Wietse Kuis, MD, PhD, Johannes W. J. Bijlsma, MD, PhD: University Medical Centre Utrecht, Utrecht, The Netherlands; 2Berent J. Prakken, MD, PhD: University Medical Centre Utrecht, Utrecht, The Netherlands and EUREKA Institute for Translational Medicine, Syracuse, Italy; 3Salvatore Albani, MD, PhD: Arizona Arthritis Center, University of Arizona, Tucson and EUREKA Institute for Translational Medicine, Syracuse, Italy. Drs. Prakken and Albani contributed equally to this work. Dr. Albani holds a patent for the peptide technology via the University of California, San Diego, for which he receives no compensation. Address correspondence and reprint requests to Berent J. Prakken, MD, PhD, Wilhelmina Children’s Hospital and University Medical Centre Utrecht, Lundlaan 6, 3584 EA Utrecht, The Netherlands. E-mail: [email protected] Submitted for publication December 7, 2007; accepted in revised form April 7, 2009. Rheumatoid arthritis (RA) is a systemic autoimmune disease that is characterized by chronic synovial inflammation of the peripheral joints (1–3). The joint inflammation in RA results from uncontrolled activation from both innate and adaptive immunity and is characterized by the infiltration of CD4⫹ T cells, macrophages, and B cells (4). Although the resulting chronic inflammation is mainly aspecific (5), the presence of large numbers of clonally expanded CD4⫹ T cells (6,7) and the association with certain HLA–DR alleles (8,9) are indicative of an ongoing antigen-driven T cell response in RA (10–13). These autoreactive T cell responses most likely target more than one antigen. Many autoantigens, including proteoglycans (14,15), type II collagen (16), gp-39 (17), p205 (18), and citrullinated 1966 IL-10–MEDIATED, PAN–DR-BINDING Hsp60–INDUCED RESPONSE IN RA proteins (19), have been proposed to play a role in the pathogenesis of RA (20). However, until now, a single causative antigen has not been identified in RA, which excludes the possibility of immune therapy through specific deletion of the autoaggressive T cells. As an alternative, epitopespecific immune therapy via the mechanism of bystander suppression or infection tolerance has been proposed (21). The major advantage of such peptide-specific immunotherapy would be that it targets specific T cells, instead of inducing nonspecific immune suppression. Ideally, epitopes for antigen-specific immune therapy should fulfill at least 3 important stipulations as follows: 1) the epitopes must be present at the site of inflammation; 2) the epitopes should be specifically up-regulated at these local sites during inflammation; 3) the epitopes must be recognized by T cells in a majority of patients, preferably irrespective of their HLA background. One of the best-studied groups of antigens that trigger T cell responses in RA is the family of heat-shock proteins (HSPs). The HSPs are highly conserved proteins that are essential for cell function. They are expressed during inflammatory conditions and are dominantly recognized by the immune system, thus fulfilling important criteria for candidate antigens for immune therapy. Indeed, a peptide derived from the DnaJ HSP has shown promise in a phase I study in patients with RA, with a phase II trial currently being performed (22,23). For the present study, we focused on another HSP, namely, Hsp60. For several reasons, Hsp60 can be considered to be a prime candidate for antigen-specific immune therapy in RA (24). First, experiments in the adjuvantinduced arthritis model suggest that Hsp60 might play a crucial role in the immune regulation of arthritis (25). Second, in RA, as well as in juvenile idiopathic arthritis (JIA), Hsp60 of microbial origins as well as those of endogenous origins are targets of immune responses (26,27). In the spontaneous remitting form of JIA (oligoarticular JIA), T cells reactive to self Hsp60 at the onset of disease are associated with disease remission (28) and have a striking regulatory phenotype (29). In RA, T cells stimulated with human Hsp60 display a marked suppressive phenotype, compared with T cells stimulated with homologous mycobacterial Hsp65 (26). Thus, Hsp60 self-reactive T cells may play a role in modulating chronic arthritis (adjuvant-induced arthritis, RA, and JIA), underscoring their potential as candidate antigens to modulate the immune response in chronic arthritis (24,30). Based on the data obtained in experimental 1967 models, it can be expected that Hsp60 will induce both proinflammatory and down-regulatory mechanisms, which may be dependent on recognition of different epitopes derived from the same protein, since, until now, T cell epitopes from Hsp60 recognized by peripheral blood mononuclear cells (PBMCs) from a broad population of RA patients have not been identified. Recently, we identified pan–DR-binding epitopes of both the human and the mycobacterial Hsp60 protein (31,32). The goal of this study was to characterize the disease specificity and function of these Hsp60-derived epitopes as contributors to the modulation of an autoimmune inflammatory loop that may be self-reverberating in a manner independent of its trigger. Our study presents novel findings of 8 pan–DR-binding Hsp60 T cell epitopes that induce a disease-specific antiinflammatory T cell response in PBMCs from RA patients, underlining the potential of these epitopes as targets for antigenspecific immunotherapy. Furthermore, these newly identified epitopes could be used as a tool to study the autoreactive T cell response in ongoing RA. PATIENTS AND METHODS Patients and cells. Heparinized blood samples were collected following the standards of the Declaration of Helsinki, from patients with RA (n ⫽ 20), patients with osteoarthritis (OA) (n ⫽ 20), and healthy control subjects (n ⫽ 20) (with the latter 2 groups of individuals being sex- and agematched to the RA patients). The patients with RA met the American College of Rheumatology (formerly, the American Rheumatism Association) 1987 revised criteria for the classification of RA (33). The patients’ characteristics are listed in Table 1. The mean duration of RA was 6.7 years (range 1–19 years). The patients with OA had been given the diagnosis in at least 1 joint and did not have infections or any immunityinterfering disease. The local medical ethics review board approved the study. PBMCs were isolated by Ficoll density-gradient centrifugation. Viable cells, checked by trypan blue exclusion, were cultured in RPMI 1640 culture medium supplemented with 100 units/ml penicillin/streptomycin, 2 mM L-glutamine (all from Invitrogen, Carlsbad, CA), and 10% heat-inactivated human AB-positive serum (Sanquin Bloodbank, Amsterdam, The Netherlands). Selection of pan–DR-binding peptides. With the use of a matrix-based computer algorithm predicting the pan–DRbinding epitopes of a given protein sequence, we selected 4 homologous pairs of peptides (p1–p8) (Table 2). Each pair consisted of a microbial Hsp60 epitope with its human analog. Mycobacterial Hsp65 was used as a matrix for microbial Hsp60. This search was based on the capacity to bind to the HLA–DR1, DR4, and DR7 binding cleft. Threshold values for the consecutive scores were ⱖ1.570 for DRB1*0101, ⱖ2.617 for DRB1*0401, and ⱖ9.106 for DRB1*0701. Peptide pairs p1/p2 and p3/p4 were chosen for the microbial epitopes (p1 1968 DE JONG ET AL Table 1. Clinical characteristics of the patients with rheumatoid arthritis (RA) and the disease and healthy control groups* Parameter Patients with established RA Patients with OA Healthy controls No. of patients Age, mean ⫾ SD (range) years Sex, no. male: no. female (% male:% female) No. (%) RF positive No. (%) with bone erosions No. (%) with rheumatoid nodules ESR, mean ⫾ SD (range) mm/hour CRP, mean ⫾ SD (range) mg/liter Disease duration, mean ⫾ SD (range) years Medication, no. (%) NSAID use DMARD use MTX use Prednisone use 20 55 ⫾ 16 (23–81) 7:13 (35:65) 10 (50) 11 (55) 4 (20) 31 ⫾ 26 (2–91) 42 ⫾ 28 (⬍5–86) 6.7 ⫾ 5 (1–19) 20 60 ⫾ 10 (42–77) 7:13 (35:65) – – – 3 (1–7)† – – 20 53 ⫾ 14 (23–75) 6:14 (30:70) – – – – – – 18 (90) 19 (95) 12 (60) 3 (15) 17 (85) 0 0 0 – – – – * The 3 groups were not significantly different from each other in age and sex distribution. OA ⫽ osteoarthritis; RF ⫽ rheumatoid factor (determined by Rose-Waaler test and by latex agglutination); ESR ⫽ erythrocyte sedimentation rate; CRP ⫽ C-reactive protein; NSAID ⫽ nonsteroidal antiinflammatory drug; DMARD ⫽ disease-modifying antirheumatic drug; MTX ⫽ methotrexate. † Tested in only 6 of the patients with OA. and p3) on the basis of an optimal pan–DR-binding score, and peptide pairs p5/p6 and p7/p8 were chosen for the pan–DRbinding score of the human epitopes (p6 and p8). The sequential use of a combined DR1/DR4/DR7 algorithm can be used to identify broadly crossreactive DR-binding peptides (34). In vitro major histocompatibility complex (MHC) binding studies have confirmed that Hsp60 peptides identified by the computer algorithm were indeed able to bind to a diverse range of HLA–DR molecules (31). The pan–DR-binding peptides were synthesized by automated, simultaneous, multiple-peptide synthesis, with a purity ⬎95%, as described previously (35). Lymphocyte proliferation assays. Cells were cultured (2 ⫻ 105 cells in 200 l per well) in triplicate in roundbottomed 96-well plates (Nunc, Roskilde, Denmark) for 120 hours at 37°C in 5% CO2 with 100% relative humidity, in the absence or presence of 20 g/ml Hsp60 peptides. Concanavalin A (2.5 g/ml; Calbiochem, San Diego, CA) and diphtheria toxoid and tetanus toxoid (5 Lf/liter; RIVM, Bilthoven, The Netherlands) were used as positive controls. A mouse class II–restricted epitope, ovalbumin (OVA) 323–339, was used as a negative control. During the last 16 hours of culture, 1 Ci 3 H-thymidine (ICN Biomedicals, Amsterdam, The Netherlands) was added to each well. Cells were harvested, and the incorporated radioactivity was measured by liquid scintillation counting, with results expressed in counts per minute. The magnitude of the proliferative response was expressed as the stimulation index (SI), which is calculated as the mean counts per minute of cells cultured with antigen divided by the mean Table 2. Characteristics of the peptides identified* Predicted binding capacity, arbitrary units Peptide/type Sequence DR1 DR4 DR7 Overall score p1/myc 254–268 p2/hum 280–294 p3 myc 216–230 myc 216–230 p4/hum 242–256 p5/myc 210–224 p6/hum 236–250 p7 myc 507–521 myc 507–521 p8/hum 535–549 myc 256–270 hum 282–296 GEALSTLVVNKIRGT GEALSTLVLNRLKVG 42.4 12.5 17.0 0.9 276.9 8.7 3 1 PYILLVSSKVSTVKD PYILLVSSKVSTVKD AYVLLSEKKISSIQS EAVLEDPYILLVSSK KCEFQDAYVLLSEKK 3.6 132.0 0.2 28.5 40.8 14.3 4.2 2.8 0.4 3.6 26.7 29.8 7.0 15.5 96.8 3 3 1 2 3 IAGLFLTTEAVVADK IAGLFLTTEAVVADK VASLLTTAEVVVTEI ALSTLVVNKIRGTFK ALSTLVLNRLKVGLQ 1.8 10.5 12.0 ND ND 0.3 1.3 3.3 ND ND 3.7 18.7 68.0 ND ND 1 2 3 ND ND * The origin and amino acid sequences of peptides p1–p8, mycobacterial (myc) Hsp65 256–270, and human (hum) Hsp60 282–296 are listed. The core sequence of each peptide is underlined. A computer algorithm predicted the core sequences of the peptides on the basis of their HLA–DR1*0101 (DR1), DR1*0401 (DR4), and DR1*0701 (DR7) binding capacity. A binding score of ⱖ1.570 for DR1*0101, ⱖ2.617 for DR1*0401, and ⱖ9.106 for DR1*0701 was assumed to be positive. Peptides with a pan–DR-1 binding score of 3 were selected for analysis, and their human and mycobacterial equivalents were also tested. ND ⫽ not determined. IL-10–MEDIATED, PAN–DR-BINDING Hsp60–INDUCED RESPONSE IN RA 1969 Figure 1. Peptide-induced T cell proliferation of peripheral blood mononuclear cells from patients with rheumatoid arthritis, disease controls (patients with osteoarthritis), and healthy controls. The results are expressed as the median stimulation index in response to peptides p1–p8, mycobacterial (m) Hsp65 256–270, human (h) Hsp60 282–296, and negative control peptide ovalbumin (ova) 323–339. Data are presented as box plots, where the boxes represent the 25th to 75th percentiles, and the lines within the boxes represent the median. Each dot represents the results for 1 patient. (The mean stimulation indices are listed in the text.) ⴱ ⫽ P ⬍ 0.05. cpm of cells cultured without antigen. If the variation between the results obtained in triplicate exceeded 2 standard deviations, the stimuli were excluded. Blocking assays. Anti–HLA–DR (clone B8.11.2) and anti–HLA class I monoclonal antibody (mAb) (clone W6/32; negative control) were obtained from Dr. F. Claas (Department of Immunohematology, Leiden University Medical Cen- tre, Leiden, The Netherlands). PBMCs (1 ⫻ 106/ml) were incubated together with these mAb against HLA–DR, for 1 hour. After this incubation, the lymphocyte proliferation assay was performed as described above. The blocking capacity of the mAb was evaluated by calculating the percent reduction in the proliferative response obtained in the blocked condition versus that obtained in the unblocked condition. 1970 DE JONG ET AL Carboxyfluorescein diacetate succinimidyl ester (CFSE) staining. PBMCs (5 ⫻ 106) were stained with CFSE (Molecular Probes, Eugene, OR) for 5 minutes in RPMI. CFSE-labeled PBMCs were then cultured in the presence of a stimulus for 168 hours. Subsequently, the PBMCs were stained for CD4 and CD3 (both from BD Biosciences, San Jose, CA), with results analyzed on a flow cytometer (FACSCalibur; BD Biosciences). Cytokine assays. For analysis of antigen-specific cytokine production, PBMCs from the patients with RA were cultured as described above. Culture supernatants were harvested after 72 hours and stored at ⫺80°C until further analyzed. Extracellular cytokines and chemokines were measured using a multiplex immunoassay (Luminex, Austin, TX) as previously described (36–38). In summary, the antibodycoated microspheres were incubated for 60 minutes with standards or culture medium (25 l) in 96-well, 1.2-m filter plates (Millipore, Amsterdam, The Netherlands). Plates were washed, and a cocktail of biotinylated detection antibodies was added for 60 minutes. After repeated washings, streptavidin– phycoerythrin was added for an additional 10 minutes. Beads were washed twice, and the fluorescence intensity was measured. Measurements and analysis of the data from all assays were performed using the Bio-Plex system in combination with Bio-Plex Manager software, version 4.1, using 5-parametric curve fitting (Bio-Rad, Hercules, CA). The concentrations of the following soluble mediators were measured: interleukin-1␣ (IL-1␣), IL-1␤, IL-2, IL-4, IL-6, IL-8, IL-10, IL-13, IL-15, IL-17, IL-23, tumor necrosis factor ␣ (TNF␣), and interferon-␥ (IFN␥). The detection range for all cytokines was 1.2–5,000 pg/ml, except IL-8 and IL-23, which had a detection range of 2.4–10,000 pg/ml (38). None of the above-mentioned cytokines is made exclusively by T cells. Therefore, we performed intracellular fluorescence-activated cell sorting (FACS) experiments to investigate which cells were producing these cytokines. Intracellular staining for IL-10, IFNy, TNF␣, and IL-4, as compared with isotype controls (all from BD Biosciences), and surface marker staining for CD3 and CD4 were performed after 48 hours of culture with the peptide of interest. The peptidespecific cytokine production was defined as the percentage of CD3⫹CD4⫹ T cells producing the cytokine of interest after epitope stimulation minus the percentage found in the nonstimulated condition and after correction for cytokine staining. Statistical analysis. For statistical analysis, SPSS software, version 10.05 (SPSS, Chicago, IL) was used. Group differences were statistically evaluated using the MannWhitney U test and the Kruskal-Wallis test. For the analysis of peptide-induced cytokine production, the geometric mean was calculated. Correlations between disease parameters and immune responses were calculated with Spearman’s test. RESULTS Recognition of pan–DR-binding epitopes in RA patients. PBMCs from 20 randomly selected RA patients (see Table 1 for demographic and clinical characteristics) were tested for their proliferative response (using 3H-thymidine incorporation) to the peptide pairs of peptides p1–p8 (Table 2). Based on observations published earlier, an SI that is twice that of the background value is defined as a positive proliferative response (31,39). The proliferative response obtained without adding the antigen (defined as the background value) ranged from 170 cpm to 1,000 cpm (mean 395 cpm). Five of the 8 peptides induced a positive T cell proliferative response in RA patients; these 5 peptides were p1 (mean ⫾ SD SI 2.5 ⫾ 0.5), p2 (mean ⫾ SD SI 4.5 ⫾ 0.8), p3 (mean ⫾ SD SI 3.3 ⫾ 0.4), p6 (mean ⫾ SD SI 2.1 ⫾ 0.4), and p8 (mean ⫾ SD SI 2.1 ⫾ 0.5) (see Figure 1 for median values; detailed results for individual patients and for each peptide tested after in vitro activation in comparison with medium activation are available upon request from the corresponding author). Interestingly, all human peptides (with the exception of p4) caused a more vigorous proliferative response in PBMCs from RA patients compared with that induced by their microbial analogs. Overall, the proliferative response to the HSP epitopes was low, which was as expected given the fact that these responses reflect a primary immune response of T cells with a low precursor frequency directed toward peptides with a high degree of homology to self. The mean ⫾ SD SI in response to whole human Hsp60 was 3.2 ⫾ 0.49, and that to the whole mycobacterial Hsp60 was 3.9 ⫾ 0.58. Notably, one of the peptides identified by the computer algorithm, p1 (mycobacterial Hsp65 254–268) (Table 2), represents only a 2–amino acid frameshift from mycobacterial Hsp65 256–270, the latter of which functions as a protective T cell epitope in the adjuvantinduced arthritis model (40). Mycobacterial Hsp65 256– 270 also induced T cell proliferation of the PBMCs from RA patients (mean ⫾ SD SI 3.9 ⫾ 0.8), while the human analog of this protective epitope, human Hsp60 282–296, was also recognized by PBMCs from RA patients (mean ⫾ SD SI 3.1 ⫾ 0.6) (Figure 1). No significant correlations were found between disease duration or the amount of inflammation and the proliferative responses. The irrelevant control peptide (OVA 323–339) did not induce significant T cell proliferation of PBMCs from RA patients. If an HSP peptide induced a positive proliferative response in vitro in an individual patient, the proliferative response was also significantly different between cultures with the HSP peptide and cultures with the control peptide (OVA 323–339) in that individual (results available upon request from the corresponding author). Thus, 5 of the 8 pan–DR-binding epitopes and 2 closely related epitopes derived from the protective epitope in the adjuvant- IL-10–MEDIATED, PAN–DR-BINDING Hsp60–INDUCED RESPONSE IN RA 1971 Figure 2. Response of peripheral blood mononuclear cells (PBMCs) from 6 randomly selected patients with rheumatoid arthritis (RA) to mycobacterial (myc) Hsp60, human (hum) Hsp60, and human peptides p1–p4, in terms of production of pro- and antiinflammatory cytokines. Production of extracellular cytokines in the supernatants of stimulated PBMCs (compared with cultures without antigen) was measured by multiplex immunoassay. A, Stimulation with the peptides resulted in a clear induction of interleukin-1␤ (IL-1␤), IL-6, IL-10, and tumor necrosis factor ␣ (TNF␣). The human peptides p2 and p4 induced fewer proinflammatory cytokines, such as IL-1␤ and TNF␣, and almost equal production of IL-10 compared with their microbial homologs. Each column represents the results from an individual patient. B, Ratio of IL-10 to TNF␣ deduced from epitope-specific cytokine production by PBMCs from 6 patients with established RA. IL-10 and TNF␣ values of 0 were replaced with a value of 0.5 to allow calculation of the ratios. Bars show the median and SEM. The human peptides p2 and p4 induced relatively more IL-10 compared with the microbial peptides, which resulted in a higher IL-10:TNF␣ ratio. induced arthritis model induced T cell proliferation of PBMCs from RA patients. Disease (RA)–specific response to the pan–DRbinding Hsp60 peptides. We next investigated whether recognition of the pan–DR-binding Hsp60 epitopes in PBMCs from RA patients is disease specific. We thus compared the results obtained from RA patients with the proliferative response of PBMCs from disease controls (patients with OA) and healthy control subjects. A positive proliferative response of the PBMCs to the pan–DR-binding peptides was observed in a higher percentage of RA patients compared with healthy controls or OA patients. Responses to the control peptide (OVA 323–331) and tetanus toxoid were not different between the 3 study groups. In supernatants cultured with p1, PBMCs from 50% of the RA patients showed a positive reaction to the peptide (SI ⬎2), compared with only 5% of the healthy controls and 20% of the OA patients. A similar pattern was seen in response to p2, with the percentages of RA patients versus healthy controls and OA patients showing a positive response being 75% versus 45% and 65%, respectively, while in response to p3, these percentages were 78% versus 45% and 50%, respectively, to p6, 62% versus 5% and 5%, respectively, to p7, 22% versus 5% and 0%, respectively, and to p8, 33% versus 0% and 0%, respectively. Similar trends were visible in response to the mycobacterial Hsp65 peptide 256–270, in which PBMCs from 65% of the RA patients showed a positive reaction compared with 25% of healthy controls and 5% of OA patients (P ⬍ 0.001 for RA versus OA). In cultures with human Hsp60 282–296, these percentages for PBMCs showing a positive response were 63% versus 40% and 10%, respectively. The differences in peptide recognition by PBMCs from RA patients and those from healthy controls reached significance (P ⬍ 0.05) for peptides p1, p2, p6, and mycobacterial Hsp65 256–270, while the difference between RA and OA patients reached significance for peptides p6, p7, mycobacterial Hsp65 256–270, and human Hsp60 282–296 (Figure 1). Thus, some of the peptides elicited proliferative T cell responses in PBMCs from RA patients but not in PBMCs from healthy controls or those from OA patients, underscoring the disease-specific recognition of these epitopes in RA. The human pan–DR-binding epitopes induced a more favorable IL-10:TNF␣ ratio compared with that induced by their microbial counterparts. PBMCs from 6 patients with RA, being representative of the whole group of RA patients, were cultured with or without the various stimuli. After 72 hours in culture supernatants, the cells were harvested, and cytokine production was measured with a multiplex immunoassay (36,38). Based on the profile of proliferative responses and the corre- 1972 sponding data on the protective epitopes in the adjuvant-induced arthritis model (p1 having close homology with a protective epitope in adjuvant-induced arthritis), we decided to focus on peptide pairs p1/p2 and p3/p4 for these further studies. As shown in Figure 2A, after culture of PBMCs from RA patients, the levels of IL-1␤, IL-2, IL-6, IL-10, IL-15, IL-17, IL-23, TNF␣, and IFN␥ were determined in the supernatants in response to both whole Hsp60 proteins and to peptides p1–p4. (Each column in Figure 2 represents the results in PBMCs from an individual patient.) We detected peptide-specific production of IL1␤, IL-6, IL-10, IL-15, and TNF␣ by PBMCs from most of the RA patients studied. The production of IL-1␣, IL-4, and IL-13 was below the limit of detection (⬍1.2 pg/ml). Production of IL-17, IL-23, and IFN␥ in supernatants with each of the epitope stimuli was similar to that in the unstimulated condition. We next investigated which cells produced these cytokines. Intracellular FACS staining showed that CD3⫹CD4⫹ T cells from RA patients produced IFN␥, IL-10, and TNF␣ after 48 hours of stimulation with peptides p1 and p2 (see Figure 3 for a representative example). The documented cytokine production by CD4⫹ T cells obviously does not exclude the possibility that other cells are also responsible for producing these cytokines upon peptide stimulation. However, we failed to detect any significant peptide-specific cytokine production by non–T cells (detailed results available upon request from the corresponding author). Taken together, these data suggest that CD4⫹ T cells are likely to be primarily responsible for peptide-specific cytokine production. Notably, when the extracellular cytokine data were analyzed further, the amount of proinflammatory cytokines (IL-1␤, TNF␣, and IL-6) produced in response to the mycobacterial peptides was elevated compared with that produced in response to their human homologs. We chose to calculate the ratio of antiinflammatory cytokines to proinflammatory cytokines. IL-10 was chosen as the antiinflammatory cytokine for this calculation, because it is known as one of the most important counterregulatory cytokines. TNF␣ was chosen as the proinflammatory cytokine for this calculation because it is the most abundant proinflammatory cytokine in RA (41). The ratios of IL-10 to TNF␣ (median ⫾ SEM) in PBMCs from an individual RA patient in response to peptides p1–p4, as shown in Figure 2B, were 0.14 ⫾ 0.02 with p1, 5.17 ⫾ 1.8 with p2, 0.16 ⫾ 0.02 with p3, and 2.02 ⫾ 0.72 with p4. Thus, the IL-10:TNF␣ ratio in DE JONG ET AL Figure 3. Peptide-specific production of interferon- ␥ (IFN ␥ ), interleukin-10 (IL-10), and tumor necrosis factor ␣ (TNF␣), but not IL-4, by CD3⫹CD4⫹ T lymphocytes. After 48 hours of culture with medium (unstimulated), with peptides p1 and p2, or with tetanus toxoid (TT) as a control, peripheral blood mononuclear cells were harvested and surface stained for CD3 and CD4, and subsequently stained intracellularly for IFN␥ and IL-10 or for TNF␣ and IL-4. Results, as assessed by flow cytometry, showed significant production of IFN␥, IL-10, and TNF␣, but not IL-4, by CD4⫹ T lymphocytes in response to peptides p1 and p2, at levels above those in the unstimulated and control conditions. response to the human peptides was more than 4 times higher than that with the microbial peptides, reflecting a more antiinflammatory phenotype. The proliferative response was caused by CD4⫹ proliferating T cells and was HLA–DR restricted. We have already shown that CD3⫹CD4⫹ T cells produced epitope-specific production of IL-10, IFN␥, and TNF␣. We next wondered whether the proliferative response to the peptides was the result of proliferating CD4⫹ T cells. Proliferating PBMC populations were monitored by fluorescent dye (CFSE) incorporation. Representative examples are shown in Figure 4. After 168 hours in culture, it was possible to identify a circumscript population of lymphocytes that had proliferated as a result of exposure to the stimuli. The lymphocytes responding to either the HSP peptides or to whole mycobacterial IL-10–MEDIATED, PAN–DR-BINDING Hsp60–INDUCED RESPONSE IN RA Figure 4. Peptide-specific proliferation of CD4⫹ T cells. Peripheral blood mononuclear cells from patients with rheumatoid arthritis were stained with carboxyfluorescein diacetate succinimidyl ester (CFSE) after 168 hours of culture in the presence of either medium, mycobacterial (myc) Hsp60, mycobacterial peptide p3, or tetanus toxoid as control. The whole lymphocyte population is indicated by the solid line, while the CD3⫹CD4⫹ lymphocyte population is indicated by the grey area. The peptide-specific lymphocyte populations proliferating in response to mycobacterial Hsp60 and mycobacterial peptide p3, in contrast to that with the control tetanus, are almost all CD4 positive. Hsp65 were almost exclusively CD4⫹ T cells. In contrast, following stimulation with tetanus toxoid, approximately one-half of the proliferating PBMCs were CD4⫹. Control experiments showed that the CD4⫹CFSElow population was indeed also CD3⫹, which was further proof of the T cell–specific proliferation. Moreover, when CD4⫹ T cells were depleted prior to stimulation, the immune response was abolished (further details are available upon request from the corresponding author). As described above, the peptides were identified by their theoretical capacity to bind to multiple HLA–DR molecules. Their pan–DR-binding capacity was also confirmed by in vitro MHC binding assays (31). To further confirm that the documented proliferative response of PBMCs from RA patients to the novel epitopes was indeed HLA–DR dependent, we performed lymphocyte stimulation assays in the presence of blocking anti–HLA–DR antibodies. For these experiments, peptides p2 and p3 were chosen, because they induced the highest proliferative responses in PBMCs from RA patients. The response to the whole human and microbial proteins could be blocked by 70–75%, while in cultures with peptides p2 and p3, an inhibition of 47% and 64%, respectively, was induced by the anti–HLA–DR mAb, resulting in normalization of the SI. Thus, the proliferation of PBMCs from RA patients, when exposed to the pan–DR-binding peptides, is a result of a class II MHC–restricted CD4⫹ T cell response. DISCUSSION The discovery of T cells with a regulatory capacity has yielded new hopes for antigen-specific immune 1973 therapy in autoimmune diseases such as RA (42). However, unlike in the experimental models, knowledge of specific antigenic targets for immune therapy in humans is still limited. HSPs are candidate antigens for such immune therapy, because they are immunodominant proteins that are up-regulated at sites of inflammation. Recently, phase I and phase II immunotherapy trials with epitopes derived from DnaJ HSP and Hsp60 showed great promise in RA (refs. 23 and 43 and Koffeman E, et al: unpublished observations) and type II diabetes mellitus (23,44), respectively. Over the years, a large amount of data has been gathered, in experimental arthritis models as well as in patients with JIA and patients with RA, suggesting that Hsp60 is a potential target for immunotherapy in arthritis (24). Whole Hsp60, however, can activate both the adaptive and the innate arms of the immune system, which underscores the importance of identifying suitable T cell epitopes for immune therapy, thus bypassing the risks of nonspecific activation. Such Hsp60 T cell epitopes would have the additional advantage that they could also be used as a tool to study the arthritis-specific T cell repertoire, which would thus lead to better understanding of the pathogenesis of RA. Although immune recognition of whole mycobacterial and whole human Hsp60 in PBMCs and synovial fluid mononuclear cells from patients with RA is well documented (26,36,45–48), no Hsp60 T cell epitope that is recognized in a majority of RA patients has yet been identified. By using a computer algorithm that can predict the potential pan–DR-binding motifs (34), we identified 8 potential T cell epitopes derived from both microbial and human Hsp60. Four of these epitopes from both microbial origin (p1 and p3) and human origin (p2 and p6) were recognized by HLA–DR– restricted CD4⫹ T cells in ⬎50% of the RA patients. The response was significantly different in comparison with that in healthy controls, for p1, p2, and p6. It is of interest that some T cell proliferation could also be detected (but to a lower extent) in PBMCs from OA patients, possibly reflecting an inflammatory process in that disease. Based on analysis of the extracellular cytokines produced by PBMCs from RA patients after exposure to the human peptides, an elevated IL-10: TNF␣ ratio was observed in response to the human peptides as compared with that in response to the microbial homologs. This proliferative response, as well as the cytokine response, was driven solely by CD4⫹ T cells. The antigen-specific responses observed (proliferation of T cells as well as elevated cytokine levels) 1974 were robust but modest, fitting a primary immune response of T cells with a low precursor frequency. Because these T cells are specific for self or almost-self antigens, it is expected that these cells are under the control of peripheral tolerance and thus do not proliferate vigorously (31,49). Of the identified epitopes, the microbial peptide p1 (mycobacterial Hsp65 254–268) and its human homolog, peptide p2, are of special interest, because the sequence of p1 is a 2–amino acid frameshift from mycobacterial Hsp65 256–270, a protective epitope in multiple models of experimental arthritis (40). Importantly, we could also document T cell recognition of this mycobacterial Hsp65 256–270 epitope in RA, as well as T cell recognition of its human homolog, human Hsp60 282–296. In experimental arthritis, this protective epitope exerts its protective effect by inducing crossreactive (self-reactive) T cells. In the RA population, we observed a correlation between the mycobacterial peptides and their human homologs for the peptide pairs p1/p2 (Spearman’s ⫽ 0.47, P ⬍ 0.05) and Hsp65 256–270/Hsp60 282–296 (Spearman’s ⫽ 0.80, P ⬍ 0.05). This may indicate that crossreactive T cells between human and microbial Hsp60 peptides are present in RA patients. Taking into account the specific recognition of both the microbial and the human epitopes, this suggests that also in RA, either p1 (mycobacterial Hsp65 254–268) or the original protective epitope identified in experimental arthritis (mycobacterial Hsp65 256–270) may be a candidate antigen for immune therapy. In order to determine the peptide recognition in a broad population of patients, we chose to perform a cross-sectional study with randomly selected patients with RA. As has been described in earlier studies, the development of responsiveness to Hsp60 is associated with disease duration (45,48). Obviously, in this study, we could not address this issue. Further research on both patients with early RA and patients with RA of longer duration should be done in order to draw any conclusions regarding the question of whether the peptide responses quantitatively or qualitatively change during the course of the disease and/or whether they may be correlated with disease outcome. One might argue that the documented recognition of several pan–DR-binding T cell epitopes in RA is a mere reflection of the known association of RA with certain HLA types, such as HLA–DR1 and DR4. However, as was shown by Southwood et al (34), the use of a combined DR1/DR4/DR7 algorithm can identify broadly crossreactive DR-binding peptides. Conse- DE JONG ET AL quently, the peptides selected by the model are able to bind to multiple HLA–DR types, as has been shown in in vitro MHC binding studies (31). To further confirm this, we performed HLA typing of the patients and related their HLA type to the proliferative response of the pan–DR-binding epitopes. As expected, no correlation was found between the HLA type and the quality or quantity of the immune response induced (results not shown). In addition, when the shared epitope (HLA– DR4) was examined in more detail, no significant differences were found when comparing the proliferative response of PBMCs obtained from DR4⫹ patients with RA with that of PBMCs obtained from DR4⫺ patients with RA. Therefore, the association between certain HLA types cannot be the explanation for the diseasespecific recognition of the pan–DR-binding epitopes in RA. Earlier studies in patients with JIA and patients with RA, using whole Hsp60 proteins, suggested that T cell recognition of human Hsp60 is associated with a regulatory phenotype (26,29,31). Interestingly, in our study, the human Hsp60 epitopes also induced a more antiinflammatory T cell response, as depicted by marked changes in the IL-10:TNF␣ ratio. This again underlines the potential role of these peptides in the ongoing inflammatory process. Whether these cells play an antiinflammatory role in vivo obviously could not be determined in this study. We previously showed that DnaJ P1, a peptide derived from another HSP, induces proinflammatory responses in PBMCs from patients with early RA and can be used to modulate autoimmune inflammation in these patients (23). The results from the present study show that the phenomenon of disease- and inflammation-specific recognition of HSP-derived epitopic peptides, which we first described in patients with RA using DnaJ P1, is, in reality, more complex and involves peptides that are functionally similar to DnaJ P1 (proinflammatory) as well as others that are quite different (50). Thus, our study contributes to a better understanding of the mechanism that can be manipulated for therapeutic purposes. The major advance is that it provides evidence of more DnaJ P1–like peptides that could be used in the same way and could, perhaps, capture a broader patient population. In addition, our findings appear to identify other peptides that probably have a different mechanism of action. Further studies are needed to evaluate the capacity of these epitopes to induce regulatory T cells in vitro and ex vivo in relation to specific disease characteristics of patients with RA. This should also yield important IL-10–MEDIATED, PAN–DR-BINDING Hsp60–INDUCED RESPONSE IN RA information on which specific epitope is most suitable for immune therapy in RA. Based on the present data, as well as the findings in experimental arthritis and the preliminary phase I study with HSP DnaJ P1 in humans, it seems that a peptide in the region of 254–270 of mycobacterial Hsp65 could be an ideal candidate for an immune therapy trial in patients with RA. 16. 17. AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Prakken had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. De Jong, Kuis, Bijlsma, Prakken, Albani. Acquisition of data. De Jong, Lafeber, de Jager, Haverkamp, Bijlsma. Analysis and interpretation of data. De Jong, de Jager, Prakken. 18. 19. 20. REFERENCES 21. 1. Feldmann M, Brennan FM, Maini RN. Rheumatoid arthritis. Cell 1996;85:307–10. 2. Sewell KL, Trentham DE. Pathogenesis of rheumatoid arthritis. Lancet 1993;341:283–6. 3. Arend WP. The pathophysiology and treatment of rheumatoid arthritis. Arthritis Rheum 1997;40:595–7. 4. Firestein GS. The T cell cometh: interplay between adaptive immunity and cytokine networks in rheumatoid arthritis. J Clin Invest 2004;114:471–4. 5. Firestein GS, Zvaifler NJ. How important are T cells in chronic rheumatoid synovitis? II. T cell–independent mechanisms from beginning to end. Arthritis Rheum 2002;46:298–308. 6. Sioud M, Kjeldsen-Kragh J, Quayle AJ, Wiker HG, Sorskaar D, Natvig JB, et al. Immune responses to 18.6 and 30-kDa mycobacterial antigens in rheumatoid patients, and V␤ usage by specific synovial T-cell lines and fresh T cells. Scand J Immunol 1991;34: 803–12. 7. Yang H, Rittner H, Weyand CM, Goronzy JJ. Aberrations in the primary T-cell receptor repertoire as a predisposition for synovial inflammation in rheumatoid arthritis. J Investig Med 1999;47: 236–45. 8. Weyand CM, Goronzy JJ. HLA polymorphisms and T cells in rheumatoid arthritis. Int Rev Immunol 1999;18:37–59. 9. Gregersen PK, Silver J, Winchester RJ. The shared epitope hypothesis: an approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum 1987;30: 1205–13. 10. Albani S, Carson DA. A multistep molecular mimicry hypothesis for the pathogenesis of rheumatoid arthritis. Immunol Today 1996;17:466–70. 11. Prakken BJ, Carson DA, Albani S. T cell repertoire formation and molecular mimicry in rheumatoid arthritis. Curr Dir Autoimmun 2001;3:51–63. 12. Cope AP, Schulze-Koops H, Aringer M. The central role of T cells in rheumatoid arthritis. Clin Exp Rheumatol 2007;25(5 Suppl 46):S4–11. 13. Isaacs JD. Therapeutic T-cell manipulation in rheumatoid arthritis: past, present and future. Rheumatology (Oxford) 2008;47: 1461–8. 14. Glant T, Csongor J, Szucs T. Immunopathologic role of proteoglycan antigens in rheumatoid joint disease. Scand J Immunol 1980;11:247–52. 15. Guerassimov A, Zhang Y, Banerjee S, Cartman A, Leroux JY, 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 1975 Rosenberg LC, et al. Cellular immunity to the G1 domain of cartilage proteoglycan aggrecan is enhanced in patients with rheumatoid arthritis but only after removal of keratan sulfate. Arthritis Rheum 1998;41:1019–25. Londei M, Savill CM, Verhoef A, Brennan F, Leech ZA, Duance V, et al. Persistence of collagen type II-specific T-cell clones in the synovial membrane of a patient with rheumatoid arthritis. Proc Natl Acad Sci U S A 1989;86:636–40. Verheijden GF, Rijnders AW, Bos E, Coenen-de Roo CJ, van Staveren CJ, Miltenburg AM, et al. Human cartilage glycoprotein-39 as a candidate autoantigen in rheumatoid arthritis. Arthritis Rheum 1997;40:1115–25. Blass S, Schumann F, Hain NA, Engel JM, Stuhlmuller B, Burmester GR. p205 is a major target of autoreactive T cells in rheumatoid arthritis. Arthritis Rheum 1999;42:971–80. Snir O, Widhe M, von Spee C, Lindberg J, Padyukov L, Lundberg K, et al. Multiple antibody reactivities to citrullinated antigens in sera from rheumatoid arthritis patients: association with HLADRB1 alleles. Ann Rheum Dis 2008. E-pub ahead of print. Cope AP, Sonderstrup G. Evaluating candidate autoantigens in rheumatoid arthritis. Springer Semin Immunopathol 1998;20: 23–39. Albani S, Prakken B. T cell epitope–specific immune therapy for rheumatic diseases. Arthritis Rheum 2006;54:19–25. Albani S, Keystone EC, Nelson JL, Ollier WE, La Cava A, Montemayor AC, et al. Positive selection in autoimmunity: abnormal immune responses to a bacterial dnaJ antigenic determinant in patients with early rheumatoid arthritis. Nat Med 1995;1: 448–52. Prakken BJ, Samodal R, Le TD, Giannoni F, Yung GP, Scavulli J, et al. Epitope-specific immunotherapy induces immune deviation of proinflammatory T cells in rheumatoid arthritis. Proc Natl Acad Sci U S A 2004;101:4228–33. Van Eden W, van der Zee R, Prakken B. Heat-shock proteins induce T-cell regulation of chronic inflammation. Nat Rev Immunol 2005;5:318–30. Van Eden W, Waksman BH. Immune regulation in adjuvantinduced arthritis: possible implications for innovative therapeutic strategies in arthritis. Arthritis Rheum 2003;48:1788–96. Van Roon JA, van Eden W, van Roy JL, Lafeber FJ, Bijlsma JW. Stimulation of suppressive T cell responses by human but not bacterial 60-kD heat-shock protein in synovial fluid of patients with rheumatoid arthritis. J Clin Invest 1997;100:459–63. Graeff-Meeder ER, van der Zee R, Rijkers GT, Schuurman HJ, Kuis W, Bijlsma JW, et al. Recognition of human 60 kD heat shock protein by mononuclear cells from patients with juvenile chronic arthritis. Lancet 1991;337:1368–72. Prakken AB, van Eden W, Rijkers GT, Kuis W, Toebes EA, Graeff-Meeder ER, et al. Autoreactivity to human heat-shock protein 60 predicts disease remission in oligoarticular juvenile rheumatoid arthritis. Arthritis Rheum 1996;39:1826–32. De Kleer IM, Kamphuis SM, Rijkers GT, Scholtens L, Gordon G, De Jager W, et al. The spontaneous remission of juvenile idiopathic arthritis is characterized by CD30⫹ T cells directed to human heat-shock protein 60 capable of producing the regulatory cytokine interleukin-10. Arthritis Rheum 2003;48:2001–10. Van Eden W, van der Zee R, Paul AG, Prakken BJ, Wendling U, Anderton SM, et al. Do heat shock proteins control the balance of T-cell regulation in inflammatory diseases? Immunol Today 1998; 19:303–7. Kamphuis S, Kuis W, De Jager W, Teklenburg G, Massa M, Gordon G, et al. Tolerogenic immune responses to novel T-cell epitopes from heat-shock protein 60 in juvenile idiopathic arthritis. Lancet 2005;366:50–6. Lehman TJ. Juvenile idiopathic arthritis and HSP60 vaccination: selective down-regulation? Lancet 2005;366:9–10. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, 1976 34. 35. 36. 37. 38. 39. 40. 41. 42. Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. Southwood S, Sidney J, Kondo A, del Guercio MF, Appella E, Hoffman S, et al. Several common HLA-DR types share largely overlapping peptide binding repertoires. J Immunol 1998;160: 3363–73. Van der Zee R, Anderton SM, Buskens C, Alonso de Valasco E, Van Eden W. Heat shock protein T cell epitopes as immunogenic carriers in subunit vaccines. In: Peptides: 1994 Proceedings of the Twenty-Third European Peptide Symposium, ESCOM. Maya HLS, editor. Leiden, The Netherlands; 1995. p. 841–2. Carson RT, Vignali DA. Simultaneous quantitation of 15 cytokines using a multiplexed flow cytometric assay. J Immunol Methods 1999;227:41–52. Khan SS, Smith MS, Reda D, Suffredini AF, McCoy JP Jr. Multiplex bead array assays for detection of soluble cytokines: comparisons of sensitivity and quantitative values among kits from multiple manufacturers. Cytometry B Clin Cytom 2004;61:35–9. De Jager W, te Velthuis H, Prakken BJ, Kuis W, Rijkers GT. Simultaneous detection of 15 human cytokines in a single sample of stimulated peripheral blood mononuclear cells. Clin Diagn Lab Immunol 2003;10:133–9. Graeff-Meeder ER, van Eden W, Rijkers GT, Prakken BJ, Kuis W, Voorhorst-Ogink MM, et al. Juvenile chronic arthritis: T cell reactivity to human HSP60 in patients with a favorable course of arthritis. J Clin Invest 1995;95:934–40. Anderton SM, van der Zee R, Prakken B, Noordzij A, van Eden W. Activation of T cells recognizing self 60-kD heat shock protein can protect against experimental arthritis. J Exp Med 1995;181: 943–52. Olsen NJ, Stein CM. New drugs for rheumatoid arthritis. N Engl J Med 2004;350:2167–79. Bluestone JA. Regulatory T-cell therapy: is it ready for the clinic? Nat Rev Immunol 2005;5:343–9. DE JONG ET AL 43. Koffeman E, Genovese M, Schiff M, Albani S. Epitope-specific immune therapy in rheumatoid arthritis: results of a phase 2 double-blind randomized clinical trial. Clinical Immunol 2006; 119(Suppl 1):S4. 44. Raz I, Elias D, Avron A, Tamir M, Metzger M, Cohen IR. Beta-cell function in new-onset type 1 diabetes and immunomodulation with a heat-shock protein peptide (DiaPep277): a randomised, double-blind, phase II trial. Lancet 2001;358:1749–53. 45. MacHt LM, Elson CJ, Kirwan JR, Gaston JS, Lamont AG, Thompson JM, et al. Relationship between disease severity and responses by blood mononuclear cells from patients with rheumatoid arthritis to human heat-shock protein 60. Immunology 2000; 99:208–14. 46. Ramage JM, Young JL, Goodall JC, Gaston JS. T cell responses to heat-shock protein 60: differential responses by CD4⫹ T cell subsets according to their expression of CD45 isotypes. J Immunol 1999;162:704–10. 47. Zou J, Rudwaleit M, Thiel A, Lauster R, Braun J, Sieper J. T cell response to human HSP60 and yersinia 19 kDa in ankylosing spondylitis and rheumatoid arthritis: no evidence for a causal role of these antigens in the pathogenesis. Ann Rheum Dis 2002;61: 473–4. 48. Ramage JM, Gaston JS. Depressed proliferative responses by peripheral blood mononuclear cells from early arthritis patients to mycobacterial heat shock protein 60. Rheumatology (Oxford) 1999;38:631–5. 49. Huurman VA, van der Meide PE, Duinkerken G, Willemen S, Cohen IR, Elias D, et al. Immunological efficacy of heat shock protein 60 peptide DiaPep277 therapy in clinical type I diabetes. Clin Exp Immunol 2008;152:488–97. 50. Massa M, Passalia M, Manzoni SM, Campanelli R, Ciardelli L, Yung GP, et al. Differential recognition of heat-shock protein dnaJ–derived epitopes by effector and Treg cells leads to modulation of inflammation in juvenile idiopathic arthritis. Arthritis Rheum 2007;56:1648–57.