Differential recognition of heat-shock protein dnaJ–derived epitopes by effector and Treg cells leads to modulation of inflammation in juvenile idiopathic arthritis

Abstract

Objective

To identify epitopes on Escherichia coli heat-shock protein (HSP) dnaJ or on homologous human HSP dnaJ involved in the induction/modulation of autoimmune inflammation in patients with oligoarticular juvenile idiopathic arthritis (JIA).

Methods

We used a proliferation assay and cytokine production to evaluate the immune responses of synovial fluid mononuclear cells (SFMCs) to pan–HLA–DR binder peptides derived from either homologous or nonhomologous regions on bacterial and human HSP dnaJ. Cytofluorometric analysis was performed in order to phenotype and sort Treg cells. Sorted cells were then analyzed for the expression of the forkhead box P3 (FoxP3) transcription factor, and their regulatory capacity was tested in coculture assays.

Results

T cell responses to E coli HSP dnaJ–derived peptides were eminently proinflammatory. Conversely, peptides derived from human HSP dnaJ induced interleukin-10 (IL-10) production from SFMCs of patients with oligoarticular JIA. A positive correlation was found between disease with a better prognosis (persistent oligoarticular JIA) and recognition of 3 human HSP dnaJ–derived peptides. The recognition of the human peptide H134–148 also induced a significantly greater amount of IL-10 in patients with persistent oligoarticular JIA than in those with extended oligoarticular JIA (P = 0.0012). Incubation of SFMCs from patients with persistent oligoarticular JIA with this human epitope increased the percentage of Treg cells and FoxP3 expression. It also induced the recovery of suppressor activity by Treg cells.

Conclusion

This is the first description of a self-regulating immune modulator circuit active during autoimmune inflammation through recognition of HSP epitopes with different functional properties. These epitopes induce T cells with regulatory function. Such induction correlates with disease severity and prognosis.

Oligoarticular juvenile idiopathic arthritis (JIA) is an example of a human autoimmune disease in which 2 subsets, each with a distinct prognosis, are present (1, 2). The extended form is characterized by progressive involvement of multiple joints, with chronic erosive disease and a worse prognosis than the self-limiting form (persistent oligoarticular JIA). Since the clinical picture is most likely affected by the underlying immune processes, oligoarticular JIA may represent a useful model for studying naturally occurring mechanisms of antigen-specific, T cell–mediated immune modulation (3–7).

One of the fundamental limitations encountered to date in studies focused on the role played by T cells in human autoimmunity is that, by analogy with animal models, research has often identified the antigen with the trigger. This approach has generated a significant amount of elegant but somewhat inconclusive data. The search for the elusive trigger of human autoimmunity could perhaps be refocused on antigens which participate in the mechanisms that are relevant to the perpetuation and modulation of autoimmune inflammation without necessarily being their trigger. This would increase the likelihood for the mechanism to be functionally relevant.

Heat-shock proteins (HSPs) are among the few families of proteins which fit all these criteria. HSPs are molecular chaperones that are highly conserved during evolution. At times of cellular stress, including infection and chronic inflammation, the expression of HSPs is up-regulated (8). It could be argued that HSPs are recognized as one of the first danger signals, and as such, they usually evoke a strong inflammatory response. Control of such responses is usually achieved through various, still poorly characterized mechanisms, which may be impaired in autoimmunity. Interestingly, it has recently been shown (9) that immunization of mice with HSP dnaJ leads to restoration of Treg cell activity, thus raising questions related to the nature of the epitope involved and to the mechanisms responsible for this effect in human autoimmunity.

We and others have previously shown that recognition of HSPs can lead to T cell responses that are immunologically relevant in autoimmune diseases (10–17). Attempts have been made to correlate such responses with the clinical picture. Such analyses did not lead to conclusive results, since often, whole proteins rather than individual epitopes were used. Identification of the appropriate epitopes is a crucial step in dissecting the molecular mechanisms of inflammation with particular focus on the identification of antigen-driven Treg cells (9, 10, 18–20). These cells play a central role in modulating antigen-specific responses and may be strongly implicated in the natural mechanisms controlling inflammation (21–24).

We studied T cell responses to epitopes derived from homologous and nonhomologous regions of Escherichia coli and human HSP dnaJ. In order to simplify the identification of the relevant epitopes among the myriad of candidates, we used a computer algorithm to scan human and bacterial dnaJ sequences for peptides bearing pan–HLA–DR binding motifs (25). We then tested the candidates for immunogenicity in patients with oligoarticular JIA and in appropriate controls. We associated the clinical picture with the quality of immune reactivity to individual peptides. We also evaluated the possible effect of recognition of these epitopes on T cells with a regulatory phenotype.

PATIENTS AND METHODS

Patients and controls.

Peripheral blood mononuclear cells (PBMCs) and/or synovial fluid mononuclear cells (SFMCs) were obtained from 45 patients (mean age 11 years, range 3–22 years) fulfilling the criteria for the diagnosis of JIA (26). Seven patients had systemic JIA, 7 had polyarticular JIA, and 31 had oligoarticular JIA. All patients with oligoarticular JIA were antinuclear antibody positive. Patients with enthesitis-related HLA–B27–positive (by serologic typing) JIA or rheumatoid factor–positive (by standard latex test) polyarticular JIA were not included in the study. At the time of sampling, all patients had active disease, defined according to the presence of synovitis upon examination; the extent of joint involvement was measured as the number of joints with active arthritis.

Patients with oligoarticular JIA were divided into 2 groups according to their disease course: those with the persistent form (n = 16), in which the disease remained limited to ≤4 joints, and those with the extended form (n = 15), in which the disease extended to involve ≥5 joints. Patients with persistent oligoarticular JIA had a stable form of the disease, as determined by at least a 1-year period of observation. Among patients with oligoarticular JIA, 4 received no treatment and all others were receiving nonsteroidal antiinflammatory drugs. Four of the patients with extended oligoarticular JIA were also receiving methotrexate. None of the patients was treated with cyclosporin A.

For controls, we used PBMCs from 10 age-matched healthy subjects who were hospitalized for minor surgical procedures. SF samples were collected at the time of intraarticular steroid injection. The outcome of the intraarticular injection was evaluated as the number of months of complete clinical response (i.e., no evidence of synovitis clinically) of the injected joint (27). Patients were routinely assessed every 3 months, or earlier if requested by parents because of disease worsening.

Consent.

The study was approved by the Institutional Ethical Committee. Permission for drawing of extra blood during routine venipuncture was obtained from the parents of all enrolled children.

Antigens.

Recombinant E coli HSP dnaJ (rdnaJ; Stressgen, Victoria, British Columbia, Canada) and 3 human HSP dnaJ (HSJ1, HDJ1, HDJ2) were used to identify class II major histocompatibility complex binder peptides.

Identification of putative epitopes.

Sequence homologies among E coli HSP dnaJ and the human homologs HSJ1, HDJ1, or HDJ2 were obtained by nonredundant scanning of the National Center for Biotechnology Information database using BLASTA and FASTA software. Prediction of HLA class II universal major binding motifs was performed using a computerized prediction of peptide binding motifs to individual HLA-derived alleles based on the BIMAS site (online at www.bimas.dcrt.nih.gov) or on pan–HLA class II binding motifs based on published algorithms (provided by AS).

The N-terminal region of E coli and of human HSP dnaJ isoforms presents homologous sequences. No sequence homology is present in the C-terminal region. Three sets of peptides were used: 1) bacterial peptides derived from both N-terminal and C-terminal regions of E coli HSP dnaJ (indicated in Figure 1 as amino acid sequences with the prefix B); 2) human peptides homologous to bacterial peptides derived from the N-terminal region of HSJ1, HDJ1, or HDJ2 dnaJ isoforms (indicated in Figure 1 as amino acid sequences with the prefix H); and 3) uniquely human peptides derived from HSJ1 and HDJ2 C-terminal regions (Figure 1). For an unrelated control peptide, we tested the 15-mer peptide pan–HLA–DR negative (AJAAAATLKAA) (J stands for cyclohexyl alanine).

Figure 1. Top, Codes and amino acid sequences of Escherichia coli heat-shock protein (HSP) dnaJ–derived peptides and their human homologs. HSJ1, HDJ1, and HDJ2 are 3 human HSP dnaJ. Bottom, Codes and amino acid sequences of uniquely human peptides.

Proliferation assay.

PBMCs or SFMCs, obtained by Lymphoprep density-gradient centrifugation (Nycomed Pharma, Oslo, Norway), were washed twice and suspended in RPMI 1640 supplemented with 10% human type AB serum (Sigma, St. Louis, MO), 2 mML-glutamine (Euroclone, Wetherby, UK), and 50 μg/ml gentamicin (Gibco, Grand Island, NY). PBMCs and SFMCs were cultured in triplicate in flat-bottomed 96-well plates (Costar, Cambridge, MA) at 3 × 10⁵/well, with or without 10 μg/ml of antigen (rdnaJ), and were incubated for 96 hours at 37°C in 5% CO₂. During the last 16 hours, ³H-thymidine was added to each well. The results are expressed as a stimulation index (mean counts per minute of cultures with antigen/mean cpm of cultures without antigen).

Generation of short-term peptide-specific T cell line.

SFMCs (3 × 10⁶) were cultured in 24-well plates (Costar) in the presence of 10 μg/ml of rdnaJ at 37°C in 5% CO₂. On day 3, cells were counted and aliquots of 3 × 10⁵ cells were incubated in 96-well plates with 2 × 10⁵ irradiated (3,500 rads) autologous SFMCs and 20 μg/ml of control peptide, peptides derived from E coli HSP dnaJ, or human HSJ1, HDJ1, or HDJ2. All experiments were set up in triplicate. Cells were cultured for 96 hours at 37°C in 5% CO₂; during the last 16 hours, ³H-thymidine was added to each well.

Cytokine measurement.

Supernatants of SFMCs were collected after 72 hours of incubation with or without bacterial or human peptides. Supernatants, which had been stored at −20°C, were used in commercially available solid-phase enzyme immunoassays for interleukin-10 (IL-10) (R&D Systems, Minneapolis, MN), interferon-γ (IFNγ) (Bender MedSystems, Vienna, Austria), or IL-4 (Ultrasensitive kit; BioSource International, Camarillo, CA).

Cytofluorometric analysis.

Freshly isolated SFMCs (5 × 10⁵) were stained with peridinin chlorophyll protein–conjugated anti-CD4 (BD PharMingen, San Diego, CA) and allophycocyanin-conjugated anti-CD25 (BD PharMingen) at 4°C for 30 minutes. The intracellular staining was performed with a Cytofix/Cytoperm kit (BD PharMingen) using phycoerythrin-conjugated anti–CTLA-4 (20 μl; BD PharMingen) and fluorescein isothiocyanate–conjugated anti–IL-10 (ImmunoKontact, AMS Biotechnology, Abingdon, UK). The same procedure was performed following a 72-hour culture of SFMCs in flat-bottomed 96-well plates in the presence of medium, the human peptide H134–148 (20 μg/ml), or the bacterial peptide B174–188 (20 μg/ml). Four hours before beginning the staining procedure, brefeldin A (10 μg/ml) was added to each well.

The samples were run on a flow cytometer (FACSCalibur; Becton Dickinson, San Jose, CA), collecting data for 2 × 10⁵ cells, and analyzed using CellQuest software (BD Biosciences, San Jose, CA). CD4+ cells with a high degree of CD25 fluorescence intensity (fluorescence intensity >10²) (region R2 [see below]) were gated and analyzed for CTLA-4 expression. CD4+,CD25+^high,CTLA-4+ cells were then gated and evaluated for intracellular IL-10. The specific percentage of positive cells was calculated by subtracting the values obtained with medium.

RNA extraction, preamplification, and complementary DNA (cDNA) synthesis.

Total RNA was isolated from CD4+,CD25+^high and CD4+,CD25^dim cells sorted (FACSVantage; Becton Dickinson) from freshly isolated SFMCs using RNeasy Micro columns (Qiagen, Valencia, CA) according to the manufacturer's instructions, adding carrier RNA at the initial step. RNA was eluted with 14 μl RNase-free water. Six microliters of RNA was preamplified using MessageAmp aRNA kit (Ambion, Austin, TX). The same procedure was performed with 5 × 10⁵ SFMCs cultured for 72 hours in flat-bottomed 96-well plates with the human peptide H134–148 (20 μg/ml) or the bacterial peptide B174–188 (20 μg/ml). CD4+,CD25+^high and CD4+,CD25^dim SFMCs were sorted at the end of the culture.

Nine microliters of the amplified RNA was used for cDNA synthesis. The oligo(dT)_12-18 (Invitrogen, Carlsbad, CA) primed first-strand cDNA synthesis was carried out in 20 μl with ImProm II reverse transcriptase (Promega, Madison, WI) according to the manufacturer's instructions (Technical Manual 236).

Real-time polymerase chain reaction (PCR).

TaqMan probes and primers were designed with Primer Express software (PE Applied Biosystems, Foster City, CA). Probes and primers were as follows: for the forkhead box P3 (FoxP3) transcription factor, forward primer 5′-TCACCTACGCCACGCTCAT-3′, reverse primer 5′-TCATTGAGTGTCCGCTGCTT-3′, probe 5′-JOE/TGGGCCATCCTGGAGGCTCCA/3BHQ1-3′; and for GAPDH, forward primer 5′-CCACCCATGGCAAATTCC-3′, reverse primer 5′-TGGGATTTCCATTGATGACAAG-3′, probe 5′-FAM/TGGCACCGTCAAGGCTGAGAACG/3BHQ-3′. The PCR reactions were carried out using TaqMan universal PCR master mix (PE Applied Biosystems) with 900 nM oligonucleotide primers (IDT Inc., Coralville, IA), 200 nM fluorogenic probe (IDT Inc.), and 5 μl of undiluted cDNA. For the final detection, the ABI Prism 7000 sequence detector (Applied Biosystems, Foster City, CA) was programmed to an initial step of 2 minutes at 50°C and 10 minutes at 95°C, followed by 45 thermal cycles of 15 seconds at 95°C and 1 minute at 60°C. Each measurement was carried out in duplicate. Calculations were done by the relative standard curve method, and results are expressed as an induction index (in arbitrary units), using the unstimulated condition in each cell population as a reference.

Coculture experiments.

SFMCs (3 × 10⁶) were cultured in 24-well plates in the presence of the human peptide H134–148 (20 μg/ml) at 37°C in 5% CO₂ for 72 hours. CD4+,CD25− and CD4+,CD25+^high SF T cells were sorted by a FACSVantage fluorescence-activated cell sorter (FACS). FACS reanalysis of an aliquot of sorted cells showed that the purity was 95% on average. FACS-sorted CD4+,CD25− T cells (1.6 × 10³) were cultured with soluble anti-CD3 (OKT-3, 30 ng/ml), the human peptide H134–148 (20 μg/ml), or the E coli HSP dnaJ–derived peptide B174–188 (20 μg/ml) in plates with 96 U-bottomed wells; similarly, 1.6 × 10³ FACS-sorted CD4+,CD25+^high T cells were cultured with soluble anti-CD3.

FACS-sorted CD4+,CD25− T cells (1.6 × 10³) were cultured in the presence of 1.6 × 10³ FACS-sorted CD4+,CD25+^high T cells with soluble anti-CD3, the human peptide H134–148, or the E coli HSP dnaJ–derived peptide B174–188 in U-bottomed 96-well plates. For a control, CD4+,CD25− T cells were cocultured with CD4+,CD25− T cells at the same ratio. Autologous irradiated (3,500 rads) PBMCs (6 × 10⁴) were added to each well as antigen-presenting cells. The cells were incubated at 37°C for 6 days, the last 16 hours in the presence of ³H-thymidine. The suppressive activity of CD4+,CD25+ cells is expressed as the percentage of inhibition of the proliferation of CD4+,CD25− cells cultured with soluble anti-CD3, the bacterial peptide, or the human peptide (mean ³H-thymidine incorporation [cpm] of triplicate wells).

Statistical analysis.

Data were analyzed using the nonparametric analysis of variance (Kruskal-Wallis test) and Duncan's test as a post hoc test. The Mann-Whitney U test for unpaired samples was used where indicated. Correlations between variables were performed by means of the nonparametric Spearman's correlation coefficient. P values less than 0.05 were considered significant.

RESULTS

Immune responses to recombinant E coli HSP dnaJ found in SFMCs of patients with oligoarticular JIA.

Proliferative responses of SFMCs from patients with oligoarticular JIA to recombinant E coli HSP dnaJ (rdnaJ) showed an SI >3 in 24 of 31 patients (range 2.1–26.5) and were significantly higher (P = 0.01) than those of the corresponding PBMCs. In contrast, SFMCs from patients with systemic or polyarticular JIA showed negligible responses that were not different from those obtained with PBMCs (Figure 2). In addition, SFMCs from patients with oligoarticular JIA cultured with rdnaJ led to a percentage of activated CD3+,CD69+ T cells significantly higher than that found in PBMCs (mean ± SEM 5.7 ± 1.0% versus 0.8 ± 0.3%; P = 0.0002).

Figure 2. Proliferation of peripheral blood mononuclear cells (PBMCs) or synovial fluid mononuclear cells (SFMCs) from healthy controls (Ctrl) or from patients with systemic (Syst), polyarticular (Poly), or oligoarticular (Oligo) juvenile idiopathic arthritis (JIA) in response to recombinant Escherichia coli heat-shock protein dnaJ. Results are expressed as the stimulation index (SI). Values are the mean and SEM. P = 0.01 for SFMCs versus PBMCs in patients with oligoarticular JIA.

Induction by E coli HSP dnaJ–derived peptides of proinflammatory T cell responses in SFMCs of patients with oligoarticular JIA.

SFMCs from patients with oligoarticular JIA were incubated with pan–HLA–DR binder peptides derived from either the N-terminal region (presenting sequence homology with human dnaJ isoforms) or the C-terminal region (presenting no sequence homology with human dnaJ isoforms) of E coli HSP dnaJ, to be tested as T cell antigens (indicated as amino acid sequences with the prefix B in Figure 1). SFMCs showed proliferative responses significantly higher (P < 0.03) than those obtained with the control peptide (pan–HLA–DR negative) following incubation with 2 N-terminal (B22–36, B174–188) and 2 C-terminal (B61–75, B268–282) E coli HSP dnaJ–derived peptides (Figure 3A).

Figure 3. A, Proliferative responses of SFMCs from patients with oligoarticular JIA (n = 31) or systemic/polyarticular JIA (n = 7 each) in the presence of control peptide (panDR−) or Escherichia coli heat-shock protein (HSP) dnaJ–derived peptides. Results are expressed as the SI. Values are the mean and SEM. ∗ = P < 0.03 versus control peptide. B, Interferon-γ (IFNγ) production in supernatants of SFMCs from patients with oligoarticular JIA (n = 31) in the presence of control peptide or of E coli HSP dnaJ–derived peptides. Values are the mean and SEM. See Patients and Methods and Figure 1 for description of peptides. See Figure 2 for other definitions.

SFMCs of patients with systemic or polyarticular JIA, tested as disease controls, showed negligible proliferative responses to bacterial peptides (Figure 3A). No significant differences in T cell reactivity were found among groups of patients undergoing different therapeutic regimens (data not shown).

The reactivity of SFMCs from patients with oligoarticular JIA to 3 of the 4 antigenic bacterial peptides was associated with a culture supernatant production of IFNγ significantly higher than that with the pan–HLA–DR–negative control peptide (Figure 3B), while IL-4 and IL-10 were not detectable (data not shown). The supernatant production of IFNγ, IL-4, or IL-10 in the presence of all the other peptides tested for the induction of proliferative responses was comparable with that obtained with the control peptide.

Human HSP dnaJ–derived peptides as a possible mechanism of naturally occurring immunomodulation in SFMCs of patients with oligoarticular JIA.

We evaluated the reactivity of SFMCs from patients with oligoarticular JIA to pan–HLA–DR binder human peptide homologs to the bacterial peptides B22–36 (H20–34, H21–35, and H23–37) (Figure 1) and B174–188 (H164–178, H167–181, and H176–190) (Figure 1) and to peptides derived from the C-terminal region of human HSP HSJ1 and HDJ2 presenting no sequence homology with E coli HSP dnaJ (Figure 1). All of the human peptides triggered cell proliferation and IFNγ production that were not significantly different from those obtained in response to the pan–HLA–DR–negative control peptide (data not shown).

In contrast to the results obtained with bacterial peptides, SFMCs from patients with oligoarticular JIA (n = 31) cultured with human peptides, homologous or not to bacterial peptides, produced detectable amounts of IL-10 (range 12–74 pg/ml) in a proportion that varied from 5% to 37% and that was higher than that obtained with the control peptide (<3.8 pg/ml). These data suggest that human-derived peptides trigger qualitatively different T cell responses that may modulate the inflammatory activity in patients with oligoarticular JIA. IL-4 was undetectable (data not shown).

We tested a limited number of patients with systemic JIA (n = 2) and polyarticular JIA (n = 3); proliferative responses to human peptides were lower than those obtained with the control pan–HLA–DR–negative peptide. In addition, cytokine production (IFNγ, IL-4, and IL-10) in response to human peptides from patients with either polyarticular or systemic JIA was always below the detection limit of the assays performed.

Association of immunomodulation by self epitopes with disease severity.

To explore the possible association between reactivity to human HSP dnaJ–derived peptides and clinical features, patients were divided according to their disease course into those with persistent (n = 16) and those with extended (n = 15) oligoarticular JIA (see Patients and Methods). As shown in Figure 4A, we found that the proliferative responses of SFMCs from patients with persistent oligoarticular JIA were significantly higher than those obtained with SFMCs from patients with extended oligoarticular JIA following incubation with peptide H167–181 or peptide H176–190, both homologs of the E coli HSP dnaJ–derived peptide B174–188.

Figure 4. A, Proliferation of SFMCs from 16 patients with persistent oligoarticular JIA (open bars) or from 15 patients with extended oligoarticular JIA (shaded bars) cultured with the H167–181, H176–190, or H134–148 human peptides. Results are expressed as the SI. Values are the mean and SEM. B, Evaluation of SFMC interleukin-10 (IL-10) production in culture supernatants. Values are the mean and SEM. P values shown are for patients with persistent oligoarticular JIA versus patients with extended oligoarticular JIA, by Mann-Whitney U test. See Patients and Methods and Figure 1 for description of peptides. See Figure 2 for other definitions.

Similar results were obtained when the uniquely human peptide H134–148 was used (Figure 4A); in addition, IL-10 production in supernatants from SFMCs stimulated with the H134–148 peptide, but not with the H167–181 or H176–190 peptides, was significantly higher (P = 0.0012) in patients with persistent oligoarticular JIA than in those with extended oligoarticular JIA (Figure 4B), indicating an important qualitative difference in T cell recognition of this peptide in persistent versus extended oligoarticular JIA. In order to further analyze possible associations with a more benign disease, we also considered the duration of the clinical remission in the studied joint following intraarticular administration of steroids as a measure of the degree of inflammatory activity in individual joints. We found that IL-10 production in response to the human peptide H134–148 was significantly directly correlated with this parameter (ρ = 0.537, P = 0.026; n = 14). The production of IFNγ in culture supernatants of SFMCs stimulated with human peptides was comparable in patients with persistent or extended oligoarticular JIA (data not shown).

Expansion of SF Treg cells in patients with persistent oligoarticular JIA resulting from recognition of the H134–148 peptide.

To determine whether the immune responses of SFMCs to the human peptide H134–148 might be related to modulation of autoimmune inflammation through the induction of Treg cells, SFMCs from 10 patients with persistent oligoarticular JIA were tested before and after 3-day in vitro culture with medium or the human peptide H134–148. B174–188 peptide, derived from E coli HSP dnaJ, was used as control. By cytofluorometric analysis (for 1 representative patient) (Figure 5A), CD4+ cells were gated according to the high degree of CD25 fluorescence intensity (region R2). CD4+,CD25+^high cells were then analyzed for CTLA-4 and IL-10 coexpression. The percentage of CD4+,CD25+^high,CTLA-4+,IL-10+ cells following incubation with the H134–148 peptide was significantly higher than that found following incubation with the B174–188 peptide (P < 0.05) and increased with respect to that found in unstimulated cells (P = 0.05) (Figure 5B).

Figure 5. A, Representative analysis of SFMCs (from 1 patient) cultured with H134–148 human peptide. SFMCs stained with anti-CD4 and anti-CD25 monoclonal antibodies were gated according to a high CD25 fluorescence intensity (region R2). B, Unstimulated CD4+,CD25+ SFMCs gated in region R2 were analyzed for the coexpression of CTLA-4 and intracellular interleukin-10 (IL-10). The same analysis was performed on these cells after a 3-day incubation with the bacterial peptide B174–188 or the human peptide H134–148. Results were obtained from 10 patients with persistent oligoarticular JIA. Values are the mean and SEM. C, CD4+,CD25+^dim (region R1 in A) and CD4+,CD25+^high (region R2 in A) SFMCs were sorted and analyzed for FoxP3 transcription factor expression in unstimulated cells and following 3 days of incubation with the bacterial heat-shock protein dnaJ B174–188 or the human peptide H134–148. Results are expressed as threshold cycle values which were normalized to GAPDH expression. The induction index (in arbitrary units [AU]) is the result of the normalization process. Values are the mean and SEM. See Patients and Methods and Figure 1 for description of peptides. See Figure 2 for other definitions.

To further corroborate the concept that recognition of H134–148 may affect Treg cell function, we tested in SFMCs from 8 patients whether in vitro exposure of SFMCs to H134–148 peptide would induce FoxP3 expression as an indication of restored Treg cell function. CD4+,CD25+^high and CD4+,CD25+^dim SFMCs (regions R2 and R1, respectively, in Figure 5A) were sorted and assessed for the expression of FoxP3 by quantitative PCR as unstimulated cells and after incubation with the human H134–148 peptide, or after incubation with the bacterial B174–188 peptide used as control. Increased amounts of FoxP3 messenger RNA (mRNA) were found in CD4+,CD25+^high cells incubated with the H134–148 peptide compared with cells incubated with the E coli HSP dnaJ–derived B174–188 (control) peptide or compared with unstimulated CD4+,CD25+^high cells (Figure 5C).

Functional CD4+,CD25+^high Treg cells induced by H134–148 peptide in SFMCs of patients with persistent oligoarticular JIA.

SFMCs from 6 patients with persistent oligoarticular JIA were cultured with the H134–148 peptide. CD4+,CD25− T cells (upper left quadrant in Figure 5A) and CD4+,CD25+^high T cells (region R2 in Figure 5A) were then sorted. The proliferative responses of the CD4+,CD25− T cells cultured with soluble anti-CD3, B174–188, or H134–148 as well as those of the CD4+,CD25+^high T cells incubated with soluble anti-CD3 are shown in Figure 6A. CD4+,CD25− T cells were then cultured in the presence of CD4+,CD25+^high T cells (at a 1:1 ratio) and soluble anti-CD3, B174–188 peptide, or H134–148 peptide. As shown in Figure 6B, CD4+,CD25+^high SF T cells did not suppress the proliferation of CD4+,CD25− T cells stimulated with soluble anti-CD3 or B174–188 peptide. In contrast, when the coculture experiments were performed with the H134–148 peptide, CD4+,CD25+^high T cells were able to suppress CD4+,CD25− responder T cells (44% mean inhibition). Suppression by CD4+,CD25+^high SF T cells was also found when CD4+,CD25− T cells were stimulated with soluble anti-CD3 in the presence of H134–148 peptide, which induced a 40% mean inhibition of the proliferation obtained in cocultures stimulated with soluble anti-CD3 alone. Due to a limited number of CD4+,CD25+^high cells, we were not able to evaluate the effect of a control peptide (i.e., B174–188) on the response to anti-CD3.

Figure 6. A, Proliferative responses of sorted CD4+,CD25− (Figure 5A, upper left quadrant) SF T cells from 6 patients with persistent oligoarticular JIA to soluble anti-CD3 (OKT3), the bacterial peptide B174–188, or the human peptide H134–148. Proliferative responses of sorted CD4+,CD25+^high SF T cells (Figure 5A, region R2) to soluble anti-CD3 (OKT3) are also shown. Results are expressed as ³H-thymidine incorporation. Values are the mean and SEM. B, CD4+,CD25− and CD4+,CD25+^high T cells were sorted from SFMCs after in vitro culture with the H134–148 human peptide. The sorted cell subsets were cocultured (at a ratio of 1:1) in the presence of soluble anti-CD3, Escherichia coli heat-shock protein dnaJ–derived B174–188 peptide, the human H134–148 peptide, or both the human H134–148 peptide and soluble anti-CD3. Results of coculture experiments are expressed as the percentage of inhibition with respect to the culture of CD4+,CD25− cells with anti-CD3, the bacterial peptide, or the human peptide. Values are the mean and SEM. See Patients and Methods and Figure 1 for description of peptides. See Figure 2 for other definitions.

DISCUSSION

We found that T cell responses to bacterial epitopes were eminently proinflammatory in nature. When we investigated responses to human peptides that were or were not homologous to bacterial peptides, proliferative responses were associated with the production of IL-10, indicating that human-derived peptides trigger T cell responses that are qualitatively different from those induced by bacterial peptides. The relative balance between these mechanisms may modulate the inflammatory activity and may be related to distinct clinical features in patients with oligoarticular JIA.

In fact, when patients with oligoarticular JIA were stratified based on their clinical characteristics and prognosis, we found that patients with persistent oligoarticular JIA had significantly greater immune responses than did patients with extended oligoarticular JIA to 2 human peptides homologous to bacterial peptide B174–188 as well as to the uniquely human peptide H134–148. This last peptide induced significantly greater IL-10 production in SFMCs from patients with persistent oligoarticular JIA than in those from patients with extended oligoarticular JIA, and this production was significantly correlated with the duration of the clinical remission in the single joints evaluated, thus suggesting a protective role for such responses. Interestingly, in vitro exposure of SFMCs from patients with persistent oligoarticular JIA to H134–148 peptide induced a percentage of CD4+,CD25+^high T cells coexpressing CTLA-4 and IL-10 higher than that obtained in SFMCs incubated with the bacterial peptide B174–188 and higher than that obtained in unstimulated SFMCs. Similarly, sorted CD4+,CD25+^high T cells showed higher amounts of FoxP3 mRNA following culture with the H134–148 peptide than following culture with the B174–188 peptide, or compared with unstimulated cells. This translated functionally into restoration of the ability of these T cells to suppress effector T cell proliferation in vitro.

Altogether, these data indicate a dichotomy in the quality of immune responses to different epitopes from the same protein. Recognition of epitopes derived from bacterial HSP dnaJ is associated with an inflammatory phenotype. Conversely, T cell recognition of self peptides is associated with immune mechanisms with regulatory function, including the presence of T cells with the regulatory functional phenotype. These responses are significantly augmented in patients with a more benign form of oligoarticular JIA, thus suggesting a direct role in modulation of autoimmune inflammation.

Recently, a number of regulatory pathways have been described in oligoarticular JIA (7, 10, 19, 20). Many studies suggest that induction of Treg cells is one of the host's natural mechanisms for controlling immune responses (i.e., inflammatory responses) (21–24). This observation was substantiated in animal models showing that the inhibition of adjuvant-induced arthritis is mediated by IL-10–driven regulatory cells induced through the nasal administration of a peptide analog of an arthritis-related Hsp60 T cell epitope (25). In addition, prevention of collagen-induced arthritis, mediated by mucosal administration of E coli heat-labile enterotoxin, involves the enhanced activity of a population of splenic CD4+,CD25+ Treg cells (28). De Kleer et al (29) showed that patients with persistent oligoarticular JIA had an increased number of CD4+,CD25+^high SFMCs compared with patients with extended oligoarticular JIA. This cell population was characterized by the expression of IL-10 and FoxP3 and by an actual suppressive function on CD4+,CD25− cells stimulated with OKT-3, therefore suggesting that CD4+,CD25+ cells may play an active role in the self-limiting and remitting character of persistent oligoarticular JIA.

These previous findings are fully consistent with our data, which show that the H134–148 self peptide incubated with SFMCs from patients with persistent oligoarticular JIA, in addition to inducing the production of IL-10, induces an increase in the percentage of CD4+,CD25+^high,CTLA-4+,IL-10+ cells and in their levels of FoxP3 mRNA. After stimulation with H134–148, these sorted cells performed an efficient suppression function when cocultured with autologous CD4+,CD25− SFMCs, while their functional regulatory capacity was totally abrogated when parallel experiments were performed with an E coli HSP dnaJ–derived peptide (B174–188), suggesting that the regulatory function of this cell population is enhanced or restored by availability of H134–148 peptide. As long as H134–148 peptide is available, such regulatory function is not limited to the same antigen; therefore, T cell clones with disparate specificity could be affected, as suggested in a different system by Thornton and Shevach (30). Interestingly, using a murine model, Nishikawa et al (9) have shown that immunization with dnaJ restores T regulatory activity, thus raising fundamental questions regarding the nature and the mechanism of action of the epitopes involved. Our research contributes by specifically addressing these questions while transferring the work from murine models to human disease.

To the best of our knowledge, this is the first study that identifies a self peptide able to enhance the function of Treg cells obtained from the inflamed joints of patients with a self-remitting form of arthritis. Although the CD4+,CD25+^high Treg cells clearly could not prevent the development of the disease, they may contribute to reversing ongoing inflammation. According to this mechanism, patients with persistent oligoarticular JIA may have partially maintained the Treg cell function in response to the H134–148 self peptide in the joint, where it is overexpressed during inflammation; this may result in the self-remitting course of the disease. This aspect may be exploited for therapeutic purposes (31). Indeed, we have recently shown in a phase I/IIa epitope-specific immunomodulation clinical trial in patients with rheumatoid arthritis that restoration of Treg cell activity is one of the main outcomes of treatment-induced immune deviation (12).

In conclusion, these data in a naturally remitting/relapsing human autoimmune disease suggest that differential recognition of epitopes from human and bacterial HSPs may be a natural mechanism for amplifying and subsequently down-regulating inflammation. This regulation may be impaired in autoimmunity and may contribute to perpetuation of the inflammatory damage. We have therefore identified a potential natural molecular regulator of inflammation that may be shared among various diseases and that we intend to exploit in future studies with the objective of identifying novel agents for immune therapy of diseases with a well-defined inflammatory component.

AUTHOR CONTRIBUTIONS

Dr. Albani 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 design. Massa, Prakken, Albani.

Acquisition of data. Massa, Passalia, Campanelli, Ciardelli, Yung, Kamphuis, Meli.

Analysis and interpretation of data. Massa, Passalia, Manzoni, Pistorio, Prakken, Martini, Albani.

Manuscript preparation. Massa, Campanelli, Sette, Prakken, Martini, Albani.

Statistical analysis. Pistorio.