IL-10 Is Critically Involved in Mycobacterial HSP70 Induced Suppression of Proteoglycan-Induced Arthritis

Background The anti-inflammatory capacity of heat shock proteins (HSP) has been demonstrated in various animal models of inflammatory diseases and in patients. However, the mechanisms underlying this anti-inflammatory capacity are poorly understood. Therefore, the possible protective potential of HSP70 and its mechanisms were studied in proteoglycan (PG) induced arthritis (PGIA), a chronic and relapsing, T cell mediated murine model of arthritis. Methodology/Principal Findings HSP70 immunization, 10 days prior to disease induction with PG, inhibited arthritis both clinically and histologically. In addition, it significantly reduced PG-specific IgG2a but not IgG1 antibody production. Furthermore, IFN-γ and IL-10 production upon in vitro restimulation with HSP70 was indicative of the induction of an HSP70-specific T cell response in HSP70 immunized mice. Remarkably, HSP70 treatment also modulated the PG-specific T cell response, as shown by the increased production of IL-10 and IFN-γ upon in vitro PG restimulation. Moreover, it increased IL-10 mRNA expression in CD4+CD25+ cells. HSP70 vaccination did not suppress arthritis in IL-10−/− mice, indicating the crucial role of IL-10 in the protective effect. Conclusions/Significance In conclusion, a single mycobacterial HSP70 immunization can suppress inflammation and tissue damage in PGIA and results in an enhanced regulatory response as shown by the antigen-specific IL-10 production. Moreover, HSP70 induced protection is critically IL-10 dependent.


Introduction
Heat shock proteins (HSP) are highly immunogenic and have the potential to trigger immuno-regulatory pathways. Although the exact mechanisms of immuno-regulation by HSP remain to be clarified, T cells specific for HSP were suggested to be involved in regulation of multiple chronic inflammatory diseases like rheumatoid arthritis (RA), diabetes and atherosclerosis [1][2][3][4].
Recent studies on HSP-responding T cells of juvenile idiopathic patients implied a regulatory role and contribution to disease remission for HSP60-specific T cells [5,6]. Among the various HSP families of molecules, the immuno-modulatory potential of HSP60 has been studied most extensively. However, also other HSP have immuno-regulatory functions. For example treatment with bacterial HSP70 has been shown to protect against development of adjuvant induced arthritis in rats, whereas treatment with other highly conserved bacterial proteins did not [7]. Exposure to bacterial HSP has been shown to activate self HSP-specific T cells that were cross reactive with bacterial HSP and induced suppression of arthritis [8,9]. In addition, a T cell line specific for an HSP70-derived peptide has been described to decrease disease severity upon transfer [10]. The protective potential of HSP, in animal models, has been reproduced by recent clinical trials with HSP-derived peptides in patients [11,12].
Even though HSP induced immuno-regulation has been studied rather extensively, the mechanism of HSP induced protection in autoimmune diseases is still not clear. A potential regulatory mechanism might be via the induction of IL-10 since the important role of IL-10 in dampening inflammation has been described extensively [13][14][15]. Additionally, IL-10 requirement for their suppressive function has been thought to be a common feature among most subsets of regulatory T cells as summarized [16].
In the present study the protective potential of mycobacterial HSP70 immunization on inflammatory disease and its dependency on IL-10 were assessed in the proteoglycan-induced arthritis model (PGIA), a progressive T cell dependent, antibody-mediated murine model for RA [17]. In this model arthritis can be induced by immunization with human proteoglycan (PG) mixed with synthetic adjuvant dimethyldioctadecylammonium bromide (DDA) instead of complete Freund's adjuvant (CFA) [18]. Therefore, the possibility of interfering with immune responses induced by mycobacterial HSP present in CFA can be excluded.
We found that HSP70 immunization dramatically suppressed arthritis development, subsequent tissue damage and the pathogenic PG-specific antibody response. Moreover, we demonstrate that HSP70 immunization results in a regulatory T cell response not only to HSP70, but also to the disease inducing PG. Finally, HSP70 did not suppress arthritis in IL-10 deficient mice, indicating that the regulatory response is IL-10 dependent.

Mice
Female BALB/c mice, retired breeders aged between 16-26 weeks, purchased from Charles River (Maastricht, The Netherlands), and IL-10 2/2 mice also on the BALB/c background (kindly provided by Dr. D. Reddick, DNAX, Palo Alto, CA) [19] were bred, housed and fed under standard conditions. Experiments were approved by the Animal Experiment Committee of Utrecht University (Utrecht, the Netherlands).

Induction and assessment of arthritis
Arthritis was induced with PG using a standard immunization protocol as described [17,18]. Briefly, 300 mg PG protein was given by intra peritoneal injection (i.p) with 2 mg of DDA (Sigma, Zwijndrecht, the Netherlands) emulsified in 200 ml PBS on days 0 and 21. PG was prepared as described elsewhere [20]. After the second PG immunization the paws of mice were examined in a blinded fashion 3 times a week to record arthritic changes of the joints. The onset and severity of arthritis were determined using a visual scoring system based on swelling and redness of paws as described [18]. Upon sacrifice, joints were fixed in 10% buffered formalin, decalcified in 0.33 M neutralized EDTA, embedded in paraffin and 5 mm sagittal sections were stained with hematoxylin and eosin.
HSP70 or control immunization HSP70 or control immunization was done, 10 days prior to arthritis induction, by intraperitoneal (i.p.) immunization with 100 mg recombinant HSP70 of Mycobacterium tuberculosis (Mt) (LIONEX Diagnostics&Therapeutics GmbH, Braunschweig, Germany), or 100 mg control protein, either recombinant enhanced green fluorescent protein (EGFP) or ovalbumin (Ova) (Sigma) in adjuvant DDA, 2 mg emulsified in PBS or with PBS in a total volume of 200 ml. To avoid interference of LPS contamination with HSP70 treatment, HSP70 containing less than 2.1 EU/mg protein was used.

Analysis of PG-and HSP70-specific serum antibody production
Antigen-specific serum antibody levels were determined by ELISA as described previously [18]. In brief, 96 well plates (Corning B.V. Live Sciences, Schiphol Rijk, the Netherlands) were coated by overnight incubation with 100 ml PG or HSP70 at 5 mg/ml in coating buffer (0.1 M NaHCO 3 , Na 2 CO 3 pH 9.6). Free binding sites were blocked with blocking buffer, Roche blocking reagents for ELISA (Roche Diagnostics, Alkmaar, the Netherlands) followed by incubation with sera at increasing dilutions and subsequently peroxidase-conjugated anti-IgG1, -IgG2a or -total IgG (BD Pharmingen, Breda, the Netherlands) in blocking buffer. After ABTS incubation serum PG-specific antibody levels were calculated as OD relative to the OD measured for the corresponding isotypes of a standard of pooled sera from arthritic mice. For HSP70-specific antibody production a reference serum of HSP70 immunized mice was used. Data were expressed relative to the average of the control group.

Measurement of antigen-specific T cell responses
Single-cell suspensions of spleens were cultured in triplicates in 96well flat bottom plates (Corning B.V.) at 2610 5 cells per well, in the presence or absence of HSP70 (10 mg/ml), PG (10 mg/ml) or Ova (10 mg/ml). Iscove's Modified Dulbecco's Medium (IMDM) supplemented with 10% FCS (Bodinco B.V., Alkmaar, the Netherlands), 2 mM L-glutamine, 100 units/ml penicillin, 100 mg/ml streptomycin, and 5610 25 M 2-mercaptoethanol was used as culture medium. After 72 or 96 hours, the cells were pulsed overnight with 3 Hthymidine (0.4 mCi per well; Amersham Biosciences Europe GmbH, Roosendaal, the Netherlands), harvested and 3 H uptake was measured by liquid scintillation counting (Microbeta, Perkin-Elmer Inc., Boston, MA). Supernatants of antigen stimulated spleen cell cultures were collected for cytokine assays after 72 hours. Supernatants harvested from cultures with cells isolated from WT mice were analyzed for IL-10 and interferon-gamma (IFN-c) expression simultaneously by multiplex analysis using the Luminex 100 system (Becton Dickinson, Mountain View, CA). The LINCOplex assay was performed according to the manufacturer's instructions (Linco Research, Inc., St. Charles, Missouri). The concentrations of IL-10 and IFN-c in the supernatants were calculated using LMAT software (Luminex Corporation, Austin, TX). IFN-c secretion by cells isolated from IL10 2/2 mice was measured by ELISA (BD OptEIA) according to manufactures protocol (BD Pharmingen).

Statistical analysis
Unless stated otherwise, data are expressed as mean6standard error of the mean (SEM). Statistical analysis was carried out using the Mann-Whitney U test (two-tailed) using Prism software (version 3.00, Graphpad Software Inc., San Diego). Significance level was set at (p,0.05).

Immunization with HSP70 decreases arthritis incidence and severity
To investigate the immuno-regulatory mechanisms of HSP70 in inflammatory diseases, the effect of mycobacterial HSP70 immunization in PGIA was studied. Mice were treated with HSP70 10 days prior to induction of arthritis by i.p. immunization with 100 mg HSP70, in the synthetic adjuvant DDA, whereas control groups received 100 mg EGFP or PBS in DDA. Subsequently, arthritis was induced by two i.p. immunizations with PG in DDA on day 0 and 21. HSP70 immunization lowered the incidence of arthritis and significantly delayed the onset of arthritis; day 5263.2 in HSP70 immunized mice compared to day 3162.0 in EGFP and 3061.0 in PBS pretreated mice (p,0.01) (Fig. 1A). Furthermore, HSP70 pre-treatment evidently reduced severity of PGIA (Fig. 1B). In association, histological analysis of joint sections of HSP70 immunized arthritic mice showed very mild leukocyte infiltration, less reactive synovial cell proliferation, and consequently almost no cartilage damage compared to joint sections of control animals ( Fig. 2A-B).

Altered antigen-specific B cell responses after HSP70 immunization
PG-specific B cell responses have been described to be required for development of severe PGIA [17]. Therefore, the PG-specific B cell response and the effect of HSP70 immunization on this response were studied. On day 38 after the first PG injection PGspecific serum IgG1 and IgG2a levels were determined by ELISA. High IgG1 levels confirmed the induction of a B cell response against the disease inducing PG in both treatment groups. However, the level of IgG2a was lowered in HSP70 compared to PBS immunized mice (Fig. 3A). In addition, we studied the pattern on days 50-64, after development of mild disease in HSP70 immunized mice. At days 50-64 the difference between PBS and HSP70 immunized mice in IgG1 levels was comparable to day 38 but the decrease in IgG2a, as observed on day 38, was no longer significant (Fig. 3B). The data indicate that mainly the PG-specific IgG2a production was influenced by HSP70 immunization and underline the importance of this isotype for induction of PGIA. Recently, HSP-specific antibodies have been shown to play a role in HSP induced regulation in models for RA [21,22]. To address whether HSP70-specific antibodies were also induced upon HSP70 immunization in the PGIA model, HSP70 serum antibody levels were analyzed by ELISA. Significantly increased HSP70-specific total IgG production was found in HSP70 immunized mice (Fig. 3C). Thus, both PG-and HSP70-specific B cell responses were altered upon HSP70 treatment as shown by a decreased pathogenic PG-specific response and an increased, possibly protective HSP70-specific B cell response.
Increased antigen-specific proliferation, IL-10 and IFN-c production after immunization with HSP70 In previous studies, HSP-specific T cell responses have been described to be involved in HSP-induced protection from disease development [7,9]. To study the effect of HSP70 immunization on differentiation of antigen-specific T cell responses, PG-and HSP70-specific proliferation and cytokine profiles were analyzed. Therefore, spleen cells from HSP70 or PBS immunized mice were harvested on day 35 after the first PG immunization. Antigenspecific proliferation was assessed by culturing the isolated cells for 96 hours with PG or HSP70 while Ova was used as control antigen. During the last 18 hours 3 H-thymidine was added and incorporation was measured and expressed as mean stimulation index (CPM Antigen/CPM background)6SEM. The mean background values (responses without antigen) were for the PBS control group 839062135 and for the HSP70 group 10 18062741. A significantly enhanced proliferative response to HSP70 restimulation was seen in HSP70 immunized mice as compared to control mice (Fig. 4A). Remarkably, also the proliferative response against the arthritis inducing PG was enhanced in the HSP70 group. Next, antigen-induced production of IL-10 and IFN-c was studied in supernatants collected after 72 hours culturing in the presence of PG or HSP70. Increased IL-10 production, by spleen cells from HSP70 immunized mice, could be detected after HSP70 in vitro restimulation (Fig. 4B). Interestingly, this increased IL-10 production was also found after PG in vitro restimulation. Furthermore, the amount of IFN-c produced by HSP70-and PG-specific cells from HSP70 treated mice was increased (Fig. 4C). In summary this clearly showed that HSP70 immunization not only induced an HSP70-specific T cell response but also modulated the T cell response to the disease inducing antigen.
Enhanced IL-10 mRNA expression in CD4 + CD25 + T cells after HSP70 immunization It has been proposed that induction or activation of HSP70specifc regulatory T cells cross-reactive with self-HSP70 are involved in protection seen after HSP70 immunization [8,9]. To study in more detail whether HSP70 immunization influenced

IL-10 is important for the protective effects of HSP70 immunization
HSP70 immunization increased the antigen-specific IL-10 production suggesting that IL-10 is important for the antiinflammatory capacity of HSP70. To address whether protection in arthritis by HSP70 immunization was indeed IL-10 dependent, IL10 2/2 mice received 100 mg HSP70 or Ova as control i.p. in DDA on day 210, followed by induction of arthritis by two injections with PG in DDA on day 0 and day 21. HSP70 immunization of IL10 2/2 mice, in contrast to wild type mice (WT), did not reduce arthritis severity as compared to control, immunized IL10 2/2 mice. In addition, there was no obvious difference in the mean maximum arthritis score or the day of onset of disease (Table 1). To study the effect of HSP70 immunization, in the absence of IL-10, on the B cell response, PG-and HSP70specific antibody production was measured in the sera of IL10 2/2 mice on day 53 after the first PG injection. Induction of arthritis resulted in a clear PG-specific antibody response as shown by elevated IgG1 and IgG2a levels in both treatment groups (Fig. 6A). However, in parallel with the arthritis scores, HSP70 immunization in IL-10 2/2 mice did not influence this PG-specific response.
As expected, HSP70 immunization enhanced the HSP70-specific IgG production (Fig. 6B). Next, induction and modulation of the antigen-specific T cell response by HSP70 treatment in IL-10 2/2 mice were addressed. Therefore, spleen cells were harvested on day 53 followed by analysis of antigen-specific proliferation after in vitro restimulation, for 72 hours with PG, Ova or HSP70. 3 H incorporation was depicted as stimulation index (CPM antigen/ CPM background). The mean background values (responses without antigen) were for the PBS control group 17056339 and for the HSP70 group 13946468. Similar to the response in WT mice, immunization of IL10 2/2 mice with HSP70 enhanced the proliferative response to HSP70 restimulation compared to control immunization with Ova. But, it did not have an effect on PGspecific proliferation (Fig. 6C). Furthermore, Ova specific proliferation was slightly increased in mice that received Ova. In addition, the effect of HSP70 immunization on antigen-specific IFN-c secretion was studied, by ELISA in the culture supernatants. Also in IL10 2/2 mice, immunization with HSP70 increased the production of IFN-c after HSP70 restimulation. But, the enhanced PG-induced IFN-c production, observed in WT mice, was not observed in IL10 2/2 mice (Fig. 6D). Taken together, this indicated that HSP70 immunization of IL-10 2/2 mice induced an HSP70-specific B and T cell response that, in the absence of IL-10, failed to suppress PGIA development and PG-specific pathogenic IgG2a production.

Discussion
The anti-inflammatory properties of HSP have been shown in several studies in both animal models and in patients suffering from inflammatory diseases [1,5,11,12,23,24]. However, for further development of HSP for therapeutic application, it is essential to understand the mechanisms by which HSP affect chronic inflammatory disease in more detail. In this study we analyzed the protective potential of HSP70 in PGIA, a chronic model for RA, studied the T cell response induced by HSP70 immunization and addressed whether the presence of IL-10 was essential. Our data clearly show that pretreatment with HSP70 delayed arthritis onset and dramatically reduced disease severity both clinically and histologically. Furthermore, it increased the antigen-specific T cell response to both the protective HSP70 and the disease inducing PG. In IL-10 deficient mice HSP70 immunization did not suppress arthritis illustrating the IL-10 dependency of HSP70 induced immuno-regulation.
The immuno-regulatory effect of HSP70 immunization has been demonstrated earlier in arthritis models in rats [9,25,26]. In contrast to the rat models, the mouse PGIA model is a chronic and progressive model of RA [27]. In addition, in this model arthritis can be induced by immunization with proteoglycan mixed with synthetic adjuvant DDA instead of CFA [18]. Therefore, the Figure 5. Increased IL-10 mRNA expression in CD4 + CD25 + cells of HSP70 immunized mice. Mice were PBS or heat shock protein 70 (HSP70) immunized before induction of arthritis as described in figure 1. On day 38, CD4 + CD25 + cells were isolated from the spleen followed by isolation of mRNA. Foxp3 and IL-10 mRNA expression was determined by Quantitative Real-Time PCR analysis. Data are normalized and expressed relative to HPRT and depicted as mean6SEM (n = 5 mice in each group). Significant difference (p,0.05, Mann-Whitney U Test) is indicated by the asterisk. doi:10.1371/journal.pone.0004186.g005 possibility of interfering immune responses induced by mycobacterial HSP present in CFA can be excluded.
In the present study an HSP70-specific T cell response was induced upon HSP70 immunization as detected by increased proliferation, IL-10 and IFN-c production upon HSP70 in vitro restimulation. Remarkably, also an enhanced PG-specific T cell response, with a similar cytokine profile, was found in HSP70 immunized mice. This might be explained by HSP70 interaction with innate receptors on antigen presenting cells, resulting in enhanced antigen priming to the subsequent PG immunization. Similarly, peptide-specific CD4 + T cell proliferation has been described to be enhanced when immunogenic peptides were fused to HSP70 subsequently resulting in increased immunogenicity of the peptides [28]. Alternatively, HSP70-induced regulatory T cells might induce a tolerogenic micro-milieu, by their cytokine production, enhancing the induction of regulatory T cells via infectious tolerance. This has been shown to occur in other models of autoimmunity in an IL-10 dependent manner [29].
Our data show that Hsp70 immunization induces a population of IL-10 producing T cells that do not express enhanced levels of Foxp3. Besides Foxp3 expressing CD25 + T cells, additional regulatory T cell subsets, such as adaptive Tr1, that do not express high levels of Foxp3 constitutively, are known [30]. Tr1 cells are induced in the periphery upon encounter of antigen and express high levels of IL-10 [31]. The HSP70-specifc IL-10 production together with enhanced IL-10 mRNA expression in CD4 + CD25 + T cells in HSP70 immunized mice suggests activation or induction of HSP70-specific regulatory T cells.
However, our observation that also CD3 2 cells produce enhanced levels of IL-10 in HSP70 immunized mice (data not shown) indicates that multiple subsets of regulatory cells can contribute to regulation and this remains to be elucidated. This is in contrast to a recent study, where we showed an increase in Foxp3 expressing regulatory T cells, after HSP treatment, in a mouse atherosclerosis model [4]. However, in the present study CD4 + CD25 + T cells from HSP70 immunized mice did not increasingly express Foxp3, indicating the possible presence of distinct types of regulation in different models of inflammation.
B cell responses play an important role in the pathogeneses of RA as demonstrated by the success of B cell targeted therapy, for example with monoclonal antibodies like Rituximab, as recently summarized [32]. Also in the PGIA model, B cells are important for disease induction [17,33]. Therefore, the effect of HSP70 immunization on the B cell response was addressed by measuring PG-specific IgG1 and IgG2a production. Interestingly, decreased PG-specific IgG2a production was found in HSP70 immunized mice, while the IgG1 levels were not significantly changed. In earlier studies the IgG1 isotype has been shown to be the dominant isotype in the PGIA model. However, in agreement with our data, IgG2a levels have been described to correlate with disease severity [34]. Furthermore, the IgG2a isotype is known to be Th1 related. In accordance with these findings PGIA is considered to be a Th1 mediated model. Thus our data indicate that the pathogenic Th1 mediated PG-specific antibody response was inhibited by HSP70 immunization. In contrast to the arthritis associated PG-specific antibody response, HSP-specific antibodies have been shown to suppress adjuvant induced arthritis in rats [21,22]. Therefore, the increased HSP70 IgG production found in our study in HSP70 immunized mice might be important for the regulatory potential of HSP70 treatment.
The increased HSP70-specifc B and T cell response, found in WT mice after HSP70 immunization, was also found in IL-10 2/2 mice. However, in the absence of IL-10 this response failed to suppress the pathogenic PG-specific B cell response and it did not increase the PG-specific T cell response as observed in WT mice. So, despite the induction of an obvious HSP70-specific response, the protective effects of HSP70 immunization on the PG-specific B and T cell response were shown to be IL-10 dependent, consequently leading to development of severe arthritis in the absence of IL-10. The important role of IL-10 in arthritis has been studied previously in the PGIA model. In this study transfer of cells expressing IL-10 after retroviral IL-10 gene transduction could suppress arthritis [35]. Moreover, IL-10 has been described to stimulate PG synthesis and to reverse cartilage degradation induced by activated mononuclear cells [14] and it negatively correlated with progression of joint destruction in RA [36], showing the important anti-inflammatory function of IL-10 in human RA.
Due to their stress inducible nature HSP can be ideal candidates for immunotherapy against chronic inflammatory diseases like RA. HSP will be upregulated, locally at the site of inflammation in the joint, as reported previously [37][38][39]. Therefore, HSP-specific immune responses can be targeted specifically to sites of inflammation. Moreover, enhanced HSP-specific T cell responses in patients have been described [5,6,40]. The observation that such responses seemed to be associated with a benign form of disease, which over time preceded disease remission [5,6], and the fact that HSP-mediated preventive and therapeutic immune interventions were effective in animal models of chronic inflammatory diseases, suggests the immunotherapeutic potential of HSP in patients with inflammatory autoimmune diseases.
In summary, this study shows that HSP70 can modulate inflammation in a model for chronic and progressive arthritis through IL-10 dependent mechanisms, operating via suppression of pathogenic antibody responses and the development of regulatory T cells. The chronic and progressive nature of the model and the possibility to use specific knock out mice, like the IL10 2/2 used in this study, will enable further mechanistic studies needed for the development of HSP based immune intervention in chronic inflammatory diseases.