Abstract
Spontaneous onset of pancreatic β cell destruction in the nonobese diabetic (NOD) mouse is preceded by the induction of autoreactive T cells, which recognize a variety of autoantigens. The 60-kDa endogenous (murine) heat shock protein 60 (hsp60) has been proposed to be one of the key autoantigens. Here we demonstrate that subcutaneous immunization of normoglycemic NOD mice with highly homologous mycobacterial or murine hsp60 activates T cells in the spleen that produce high levels of IL-10 upon restimulation in vitro with either hsp60 protein. In time, increasing levels of hsp60-induced IL-10 could be detected in NOD mice, but not in age- and MHC class II-matched BiozziABH mice, which lack any sign of pancreatic inflammation. These results suggest that the IL-10 responses in NOD mice are primarily driven by endogenous inflammation. Genetically protected NOD-asp mice, showing a less progressive development of insulitis, demonstrated a similar increase in hsp60-induced IL-10 in time compared with wild-type NOD mice. Taken together, our results suggest that endogenous hsp60 is not a primary autoantigen in diabetes but is possibly associated with regulation of insulitis. Moreover, the capacity to respond to (self) hsp60 is independent of the MHC class II-associated genetic predisposition to diabetes.
Insulin-dependent diabetes mellitus (IDDM)3 is an autoimmune disease resulting from T cell-mediated autoimmune destruction of the insulin-producing β cells in the pancreatic islets of Langerhans. Among several candidate autoantigens, endogenous heat shock protein (hsp) 60 is suggested to be involved in this process, although strong evidence to support this hypothesis in IDDM patients is lacking (1, 2). In the spontaneous nonobese diabetic (NOD) mouse model for IDDM, it was indeed shown that naive NOD T cells could be triggered by murine (self) hsp60 (m60), as well as by mycobacterial (foreign) hsp60 (Mt60) (3). However, Tisch and coworkers demonstrated that cellular and Ab responses to m60 in NOD mice were not detected until 8 wk of age, whereas other autoantigens, e.g., GAD65, elicited responses as early as 4 wk of age (4). These data argue against a critical role for hsp60 as a primary autoantigen in experimental diabetes.
Mt60 shares a high degree of amino acid identity with the mammalian homologues (5). Therefore, immunization with Mt60 could result in the specific activation of cross-reactive T cells, recognizing conserved hsp60 epitopes. From experimental arthritis models in the rat, it has become clear that immunization with Mt60 results in the induction of arthritis-regulatory T cells (6), which respond to a conserved part of hsp60 present in both the microbial and mammalian protein (7). Despite the fact that cross-reactive T cell recognition is a well-accepted concept for the induction of autoimmunity (molecular mimicry), we hypothesize that conserved epitopes of hsps are involved in the maintenance of peripheral tolerance and regulation of inflammation via the selective induction of regulatory T cells (5). However, the phenotype and regulatory mechanism(s) of self hsp60-responding T cells remain to be elucidated.
Several reports by Cohen and coworkers have demonstrated that vaccination with either the complete Mt60 protein or two particular epitopes of human or m60, p277 and p12, can arrest diabetes development in the NOD model (8, 9, 10). On the contrary, p277-specific T cell clones were capable of inducing diabetes (11). These results suggest that a certain episode of acute autoimmunity may activate specific regulatory mechanisms that subsequently cope with the chronic autoimmune diabetogenic process (12).
To study our hypothesis that conserved hsp60 epitopes could be involved in the regulation of inflammation such as insulitis (5), we analyzed proliferative and cytokine responses to mycobacterial and self (murine) hsp60 upon immunizing normoglycemic female NOD mice with either hsp60 protein. Results from age-matched diabetes-susceptible NOD mice were compared with results obtained from genetically protected BiozziABH and MHC transgenic NOD mice. In the latter strain, the gene encoding the Aβ has been mutated at position 57, in which the serine was changed into an aspartate, resulting in the coexpression of wild-type I-Ag7 and I-Ag7asp (NOD-asp). As a result of this transgene, these mice have a marked reduction in spontaneous diabetes incidence, despite the presence of cellular infiltrates in the pancreatic islets, which is not affected by cyclosphosphamide treatment (13). The importance of an aspartate at position 57 of the HLA-DQβ chain (the human homologue of the murine I-Aβ chain) in determination of IDDM susceptibility was demonstrated by Todd and colleagues (14). They suggested that the amino acid at this position affects the overall structure of the DQβ chain and the conformation of the peptide binding groove. Consequently, the expression of the I-Ag7asp transgene could have pronounced effects either on thymic negative selection by eliminating autoaggressive T cells (13, 15, 16) or on positive selection of regulatory T cells, which subsequently inhibit diabetogenic T cells selected via I-Ag7 in the periphery (17, 18). Alternatively, the peptide pool presented in the periphery during an autoimmune response could be altered, thereby affecting the phenotype of diabetogenic T cells (19). Because genetic factors determining the capacity to respond to self proteins are unknown, we evaluated I-Ag7-restricted hsp60-specific T cell responses in the presence or absence of insulitis, as well as in the presence or absence of the protective I-Ag7asp molecule.
Materials and Methods
Mice
Wild-type NOD (NOD-wt)/LtJ mice were either derived from the Department of Immunology, Erasmus Universiteit Rotterdam, The Netherlands (NOD/LtJ/eur), or from the Department of Pathology, Cambridge University (NOD/LtJ/cam); both lines were bred from a breeding nucleus provided by Dr. E. Leiter (The Jackson Laboratory, Bar Harbor, ME). Transgenic NOD-asp mice were generated as described before (13) and bred in the Department of Pathology, Cambridge University. BiozziABH/RijHsd mice were obtained from Harlan (Horst, The Netherlands). The mice were bred and kept under specific pathogen-free conditions in filter-top cages with free access to acidified water and irradiated food pellets. Female mice were used in the experiments, which were performed in compliance with the guidelines of the Animal Ethics Committee of the Vrije Universiteit Amsterdam.
Ags and immunizations
Purified recombinant hsp60 of Mycobacterium bovis bacillus Calmette-Guérin (identical with hsp60 of Mycobacterium tuberculosis) was generously provided by Dr. R. van der Zee, Utrecht University, The Netherlands (7). Escherichia coli bacteria transformed with a full-length m60 expression plasmid were obtained from Dr. R. Tisch, University of North Carolina, Chapel Hill, NC (4). Recombinant m60 was purified using the QIAexpress Ni-NTA protein purification system (Qiagen via Westburg, Leusden, The Netherlands). Hen egg lysozyme (HEL) was used as control Ag (Sigma via Brunschwig Chemie, Amsterdam, The Netherlands).
For immunization experiments, Ags were mixed in a 1:1 ratio with dimethyl dioctadecyl ammonium bromide (DDA; Eastman Kodak, Rochester, NY), which was prepared as a 20 mg/ml gel in sterile PBS and used as adjuvant (20). Mice were immunized s.c. in the base of the tail with 100 μg recombinant hsp60 or HEL protein mixed with DDA adjuvant. Spleens were collected 10 to 14 days later and cells were used after preparing a single-cell suspension to determine T cell responses in vitro.
T cell proliferation assays
To induce proliferative responses, splenocytes were cultured in triplicate in a final volume of 200 μl/well in 96-well flat-bottom microtiter plates (Costar, Cambridge, MA) at 2 × 105 cells/well in the presence of Ag or medium alone. For induction of cytokines, 5 × 106 cells were cultured in a final volume of 0.5 ml/well in 24-well plates. In all assays, cells were cultured in IMDM supplemented with 5–10% heat-inactivated FCS, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin (all from Life Technologies, Breda, The Netherlands), and 5 × 10−5 M 2-ME. Splenocytes were tested for proliferative responses to recombinant hsp60 proteins at varying doses as indicated in figure legends. As a positive control for T cell proliferation, mitogenic stimulation with Con A (2.5 μg/ml final concentration) was performed. To exclude the possibility that the in vitro responses resulted from mitogenic stimulation by possibly contaminating levels of LPS in the recombinant proteins, parallel spleen cell cultures were done in the presence of 1 μg/ml LPS, which did not exceed results from cultures done in medium alone (data not shown). For cytokine induction, cells were stimulated with either 5 μg/ml ConA or 100 μg/ml Mt60 or m60. Cultures were incubated for 96 (proliferation) or 72 (cytokines) hours at 37°C in a humidified atmosphere containing 5% CO2. Proliferation cultures were pulsed for the final 16–20 h with 0.4 μCi/well [3H]TdR with a specific activity of 1 Ci/mmol (Amersham International, Bucks, U.K.). TdR uptake was measured using a liquid scintillation beta counter, and results are expressed as mean cpm. Responses to Ag were compared with responses in cultures performed in the presence of medium alone. Proliferative responses demonstrating a stimulation index (SI) ≥2 were considered to be positive. From parallel cytokine cultures, supernatants were harvested and stored at –70°C until further analysis.
Cytokine measurements
To determine cytokine levels in culture supernatants, Nunc Maxisorb plates (Nunc, Roskilde, Denmark) were coated overnight at 4°C with 5 μg/ml anti-mouse IFN-γ (XMG1.2), 1 μg/ml anti-mouse IL-10 (SXC1.1), or 2 μg/ml anti-mouse IL-4 (11B11), each diluted in PBS (21). Subsequently, nonspecific protein binding was prevented by incubating the plates for 30 min at room temperature with a 1% BSA blocking solution in PBS. After washing the plates with PBS containing 0.05% Tween 20 (PBT), supernatants diluted in IMDM were incubated for 2–3 h at room temperature. After washing with PBT, biotinylated anti-mouse IFN-γ (R46.A2, 2 μg/ml), anti-mouse IL-10 (JES5-2A5, 0.5 μg/ml), or anti-mouse IL-4 (BVD6-24G2, 0.5 μg/ml) was added for an additional hour, followed by washing with PBT and addition of 1.6 μg/ml peroxidase-conjugated streptavidin (Dako, Glostrup, Denmark). Abs were diluted in PBT. Finally, the plates were washed thoroughly with PBT, and 2 mg/ml O-phenylenediamine-dihydrochloride (Sigma) in 0.1 M phosphate-citrate buffer containing 0.015% hydrogen peroxide was used for color development. The plates were read at 490 nm, and cytokine concentrations were determined with reference to a standard curve constructed using serial dilutions of recombinant murine IFN-γ, IL-10, or IL-4 (all from Genzyme via Sanbio, Uden, The Netherlands). The anti-cytokine Ab-producing hybridomas were kindly provided by Dr. R.L. Coffman (DNAX, Palo Alto, CA).
Monitoring for development of diabetes
Mice were monitored weekly for clinical signs of diabetes using Gluko-Test reagent sticks, detecting the presence of glucose in the urine (Boehringer Mannheim, Almere, The Netherlands). On some occasions, blood glucose levels were determined using Haemo-Glukotest 1-44R reagent sticks and a Reflolux S glucometer (Boehringer Mannheim). Mice were considered diabetic when urine glucose levels exceeded 55 mmol/l or when plasma glucose levels exceeded 11 mmol/L on 3 different days within 1 wk.
Histology
Pancreatic tissue was snap frozen in liquid nitrogen and stored at −70°C until preparation of 8-μm cryostat sections, which were fixed in 10% formalin and subsequently stained with hematoxylin and eosin for the detection of insulitis in individual islets.
Statistical analysis
After log transformation, data were analyzed via two-way ANOVA and Student’s t test, using SPSS 5.0 software. Values of p < 0.05 were considered significant.
Results
Immunization with Mt60 or m60 selectively activates T cells that produce high levels of IL-10 upon recognition of either hsp60 protein
To investigate whether the T cell repertoire of NOD mice contains T cells recognizing both m60 (self hsp60) and Mt60, we determined proliferative and cytokine responses upon immunization of normoglycemic female NOD mice with either Mt60 or m60. Subcutaneous immunization with Mt60 (Fig. 1⇓) resulted in proliferative responses to both Mt60 (SI 9.8) and m60 (SI 5.9) in draining (inguinal) lymph node cultures (Fig. 1⇓, lower left). Splenocyte cultures (Fig. 1⇓, upper left) demonstrated very low proliferative responses to either Mt60 (SI 2.0) or m60 (SI 3.1). In contrast, immunization of NOD mice with m60 induced clear recall responses to m60 by both spleen (Fig. 1⇓, upper right, SI 5.5) and inguinal lymph node (Fig. 1⇓, lower right, SI 16.8) cells, whereas the mycobacterial homologue induced much lower proliferative recall responses (SI 2.0 for spleen and SI 3.5 for lymph node cells).
Analysis of T cell cross-reactivity after immunization with Mt60 or m60. Normoglycemic 11-wk-old female NOD mice (n = 5) were immunized in the base of the tail with either 100 μg recombinant Mt60 (left graphs) or m60 (right graphs) mixed with DDA adjuvant. Ten days later, spleens and draining inguinal lymph nodes were collected, pooled, and tested in vitro for recall responses to m60 (•) or Mt60 (○). Control stimulation was done in medium alone (med). Results are shown as mean proliferative response ± SD of triplicate cultures.
We next analyzed corresponding cytokine levels in spleen and inguinal lymph node cultures of mice immunized with either hsp60 protein (Fig. 2⇓). IFN-γ levels were most pronounced after restimulation of inguinal lymph node cells (Fig. 2⇓, ▪) with the hsp60 protein, which was used for immunization. However, the homologue protein failed to induce similar levels of IFN-γ. Splenic IFN-γ levels (Fig. 2⇓, ▨) displayed the same recall-response pattern, but the overall levels measured were ∼3-fold lower than measured in the lymph nodes. In contrast to lymph node cells, spleen cells derived from either Mt60- or m60-immunized mice displayed high levels of IL-10 in response to both hsp60 proteins. Collectively, these results suggest that immunization with (self) hsp60 activates T cells in the spleen that produce high levels of IL-10 but low IFN-γ upon recognition of conserved determinants on both Mt60 and m60.
Corresponding cytokine responses upon Mt60 or m60 immunization. Pooled spleen cells (▨) or inguinal lymph node cells (▪) were tested for IFN-γ (top) or IL-10 (bottom) production in response to 100 μg/ml Mt60, m60, or medium alone. Supernatants were tested for the presence of cytokines via ELISA as described in Materials and Methods. Results are shown as mean cytokine levels ± SD of quadruplicate cultures.
T cell reactivity to hsp60 in time is characterized by decreased proliferative responses but increased IL-10 responses
Considering the typical anti-inflammatory splenic cytokine profile observed upon immunizing prediabetic NOD mice with hsp60, we performed a cohort analysis on hsp60-specific proliferation and cytokine production (Fig. 3⇓) by spleen cells from normoglycemic mice immunized with Mt60 at various ages before onset of diabetes. At 5 wk of age, when the first signs of perivascular infiltration are observed (data not shown), Mt60 elicited clear dose-dependent proliferative responses (Fig. 3⇓, top). At 11 wk of age, when most of the islets demonstrate peri-islet infiltration, significant proliferation was observed only after stimulation with the highest concentration of Mt60 (data not shown). At 14 and 17 wk of age, when an increasing number of islets is fully infiltrated, significant proliferative responses to Mt60 could not be detected, despite normal responses to mitogenic stimulation (ConA). At 5 wk of age, low levels of hsp60-induced IFN-γ were measured, which increased in time (Fig. 3⇓, center). At all time points tested, hsp60-specific IL-4 levels were below the detection limit of our ELISA (data not shown). IL-10 levels induced by Mt60 demonstrated a clear significant increase between 5 and 14 wk of age (Fig. 3⇓, bottom). Interestingly, analysis of IL-10 responses in cultures derived from normoglycemic mice immunized at 17 wk of age demonstrated significantly lower levels of IL-10 when compared with earlier time points. Around this time point, 30–50% of the female NOD mice in our NOD colony became overtly diabetic, indicating that the onset of diabetes was accompanied by a reduction of hsp60-specific IL-10 production.
Splenic Mt60-specific responses induced at different stages during development of insulitis. A cohort of normoglycemic female NOD mice was immunized with Mt60 at 5 (▪), 11 (▩), 14 (▨), or 17 wk (□) of age, respectively. To determine proliferative responses (top), total spleen cells were collected and stimulated in vitro in the presence of 10 μg/ml Mt60, 2.5 μg/ml ConA, or medium alone. Supernatants were collected from parallel cultures and tested for IFN-γ (middle) or IL-10 (bottom). Data are shown as mean proliferative responses or cytokine levels ± SD of four of five spleens per group; ∗, p < 0.05 compared with medium-stimulated cultures at the same time point. Δ, p < 0.05 compared with Mt60 stimulation at 5 wk of age.
Development of insulitis is followed by increased levels of hsp60-induced IL-10
Comparison of MHC class II-matched BiozziABH and NOD mice.
Based on the observed kinetics and cross-reactive nature of hsp60-induced IL-10 production, elevated IL-10 levels could result from increased presentation of self hsp60 during ongoing inflammation (22). Therefore, we analyzed IL-10 responses in age-matched BiozziABH mice (Table I⇓), which express the same I-Ag7 MHC class II molecule as NOD-wt mice but lack any sign of insulitis or diabetes (23). Groups of age-matched, female BiozziABH and NOD mice were immunized with Mt60 or control HEL protein at 11 wk of age, at which time point NOD mice displayed clear signs of peri-islet infiltration. Mt60-restimulated splenocytes from Mt60-immunized BiozziABH mice produced significantly less IL-10 when compared with NOD mice (p < 0.05), whereas IFN-γ production was similar in both strains. Again, IL-4 levels were below the detection limit of our ELISA (data not shown). In addition, we determined Mt60-induced cytokine levels in spleen cell cultures derived from HEL-immunized BiozziABH and NOD mice. NOD spleen cells were found to produce significantly higher levels of cytokines, in particular IL-10, whereas Mt60 stimulation hardly elicited cytokine production by BiozziABH splenocytes. Because HEL-immunized NOD mice displayed equal levels of HEL-specific IL-10 in time when compared with HEL-immunized Biozzi mice, our data point out that the observed elevated IL-10 response in prediabetic NOD mice is specific for hsp60 only. Moreover, these results demonstrate that priming of hsp60-specific IL-10-producing cells occurs spontaneously in NOD mice and is affected by additional subcutaneous immunization.
Cytokine responses in MHC and age-matched BiozziABH and NOD mice
Comparison of susceptible NOD and genetically protected NOD-asp mice.
An alternative explanation for the observed difference in IL-10 production could be the different genetic background of the MHC class II-matched NOD and BiozziABH mice. Therefore, we analyzed development of insulitis and hsp60 T cell responses in NOD-asp transgenic mice, which have the same genetic background as the NOD mice, including the expression of I-Ag7, but are almost completely protected from spontaneous diabetes development due to additional expression of an I-Ag7asp transgene (13). For these studies, we used NOD-wt mice derived from the same breeding center (NOD/LtJ/cam) for comparison, as the course of insulitis and subsequent diabetes development is known to differ significantly between the various NOD colonies (24). At 32 wk of age, female NOD-asp mice show a marked decrease in spontaneous diabetes incidence (10–12%) when compared with wild-type (NOD-wt) mice (75%) from the Cambridge colony (Fig. 4⇓). In addition, insulitis development is delayed in transgenic NOD-asp mice, as demonstrated by the photographs of pancreatic sections taken at different time points. At 11 wk of age (Fig. 4⇓, a and d), the majority of islets in NOD-wt mice displayed peri-islet infiltrates, whereas most islets of NOD-asp transgenic mice were completely devoid of cellular infiltrates (Table II⇓). Only a small number of islets showed minor peri-islet infiltration. At 15 wk of age (Fig. 4⇓, b and e), most islets of NOD-wt mice demonstrated major peri-islet infiltration, whereas 45% of NOD-asp islets displayed minor peri-islet infiltration. At 18 wk of age (Fig. 4⇓, c and f), the majority of islets in NOD-wt mice demonstrated full islet infiltration, whereas NOD-asp islets displayed minor peri-islet infiltrates, comparable with the grade of insulitis observed at 15 wk of age.
Insulitis and spontaneous diabetes in n = 20 female NOD-wt (NOD/LtJ/cam) and transgenic (NOD-asp) mice. Corresponding photographs show cryostat sections of pancreatic tissue stained with hematoxylin/eosin at 11 (a and d), 15 (b and e), and 18 (c and f) wk of age, respectively (magnification, ×400).
Development of insulitis in NOD-wt and -asp micea
Based on the difference in insulitis development and spontaneous onset of diabetes between NOD-wt and NOD-asp mice, we studied whether these differences could be correlated to differences in hsp60 reactivity. Therefore, we measured hsp60-specific proliferative and cytokine responses in spleen cell cultures from age-matched normoglycemic wild-type and NOD-asp mice immunized with Mt60 at 11, 15, and 18 wk of age (Figs. 5⇓ and 6⇓). NOD-wt mice that became diabetic before 18 wk of age were excluded from the experiments. As shown before in NOD mice from the Rotterdam NOD colony (Fig. 1⇑, upper left), splenocytes from Mt60-immunized NOD-wt mice demonstrated a dose-dependent, low proliferative response to either Mt60 or m60 at all time points (Fig. 5⇓). Comparable with NOD-wt mice, transgenic NOD-asp mice responded to Mt60 at 11 and 15 wk of age, whereas the proliferative response tended to be lower at 18 wk of age. Interestingly, proliferative responses to m60 could not be detected at 11 wk of age, whereas at 15 wk of age the response was comparable with that of NOD-wt mice. Again, the response to m60 tended to be lower at 18 wk of age.
Comparison of proliferative responses to hsp60 between NOD-wt and transgenic (NOD-asp) mice. Age-matched (n = 5), normoglycemic mice from both strains were immunized in the base of the tail with 100 μg Mt60 mixed with DDA adjuvant at 11, 15, and 18 wk of age, respectively. Ten days later, spleens were collected, pooled, and tested in vitro for proliferation in response to Mt60 (○) or m60 (•) at different concentrations (μg/ml) as indicated on the horizontal axis. Results of triplicate cultures are shown as mean proliferative response ± SD.
Comparison of IFN-γ and IL-10 production between NOD-wt and transgenic (NOD-asp) mice upon subcutaneous immunization with Mt60. NOD-wt (□) or transgenic NOD (▪) mice were immunized in the base of the tail at 11, 15, or 18 wk of age. Ten days later, spleens were collected, pooled, and tested in vitro for IFN-γ (top graphs) and IL-10 production (bottom graphs) when restimulated in the presence of 100 μg/ml mycobacterial hsp60 (left graphs) or m60 (right graphs). Results of quadruple cultures are shown as mean level ± SD.
IFN-γ and IL-10 levels were determined in parallel spleen cell cultures from either NOD-wt or NOD-asp mice (Fig. 6⇑). In time, a clear decrease in IFN-γ production in response to Mt60 was observed, which coincided with a marked increase in IL-10 production. Stimulation with m60 induced only low levels of IFN-γ, which hardly varied in time. Except for IL-10 levels measured at 15 wk of age, both hsp60 proteins induced comparable IL-10 levels by either mouse strain despite major differences in pancreatic insulitis (Table II⇑). The highest levels of IL-10 were observed at 18 wk of age. Interestingly, none of the NOD-asp and only 20% of the NOD-wt mice demonstrates hyperglycemia at this time point (Fig. 4⇑), suggesting a correlation between high levels of hsp60-specific IL-10 and successful regulation of insulitis in these mice.
Discussion
In the present study, we show that subcutaneous immunization of prediabetic NOD mice with either Mt60 or m60 activates T cells that upon recognition of either hsp60 protein produce low levels of IFN-γ but high amounts of IL-10, a cytokine well known for its anti-inflammatory properties (25, 26, 27, 28). Interestingly, several murine studies have demonstrated that this particular cytokine profile is associated with regulation of immune responses (29, 30, 31), although human T cells producing IL-10 in the context of IFN-γ are sometimes regarded as proinflammatory (32, 33). Because the IL-10 and IFN-γ production could be blocked in vitro by anti-CD4 and anti-MHC class II Abs (data not shown), hsp60-specific T cells produce IL-10 either directly or indirectly by activating APC to produce IL-10. Regulatory IL-10-producing T cells (Tr1) and IL-10-producing APC have both been reported to suppress Th1 responses (29, 34), suggesting a role for Tr1 cells and/or IL-10-producing APC in regulation of autoimmune phenomena such as insulitis.
Considering the insulitis-related increase in IL-10 production upon subcutaneous immunization, one could hypothesize that increased expression of endogenous hsp60 during progressive insulitis is driving regulatory cytokine responses. Therefore, exogenously provided cross-reactive epitopes presented upon immunization with Mt60 could enhance the T cell response, which was initiated by endogenously presented self hsp60 epitopes. Spontaneous development of insulitis is known to coincide with elevated levels of endogenous hsp60 in the cytoplasm of β cells (35) as well as β cell apoptosis (36). It is conceivable that upon phagocytosis of apoptotic β cells and migration to pancreatic draining lymph nodes, APC prime endogenous hsp60-specific T cells, which subsequently migrate to the spleen and pancreatic islets. This would account for the presence of IL-10-producing hsp60-specific T cells in the spleens of either naive or Mt60-immunized mice, but not in peripheral lymph nodes. On the contrary, high IFN-γ-producing T cells were found only in the lymph nodes draining the site of immunization. Proinflammatory conditions, as induced by using DDA adjuvant, favor the generation of high IFN-γ-producing T cells, perhaps by altering the selective processing and/or presentation of conserved hsp60 epitopes. Analysis of cytokines produced by pancreas-infiltrating hsp60-specific lymphocytes at different stages during insulitis development would be helpful to elucidate whether the IL-10-producing T cells as identified in the spleen are indeed involved in local regulation of insulitis. Results reported by Birk et al. (37) on transgenic NOD mice, expressing an m60 transgene under the H-2Eα class II promotor, demonstrated substantially reduced insulitis, rarely progressing beyond the stage of peri-islet infiltration. These observations indeed suggest an important role for hsp60 in preventing complete infiltration of the islets. In addition, Kallmann et al. (38) recently demonstrated that discordant twins without signs of autoimmunity (e.g., islet cell Ab negative) produced significantly higher levels of IL-10 upon stimulation with human hsp60 when compared with their diabetic cotwins, underlining a potential role of hsp60 in regulation of type I diabetes in humans.
The remarkable differences in hsp60-induced IL-10 production and progression to overt diabetes around 18 wk of age between NOD/LtJ mice from two separate colonies suggest that successful maintenance of immune regulation modulates pancreatic inflammation and delays subsequent diabetes development. Inability to maintain high levels of IL-10 could be indicative of spontaneous alterations in the delicate balance between autoaggressive and immune-regulatory T cells as described previously (39). Intriguingly, we observed a correlation between IL-10 production and extent of insulitis at 18 wk of age. High production of IL-10 in the spleen was found to be associated with less affected pancreatic islets and slow progression to overt diabetes, whereas low IL-10 production was correlated with extensive insulitis and rapid onset of overt diabetes. These differences in disease kinetics in the two colonies might explain our failure to delay the rapid diabetes onset in the latter NOD mice by immunizing them with a single injection of Mt60 as described by Cohen and collegues (8).
The hsp60-specific IL-10 levels are not uniquely observed in the NOD mouse, as demonstrated by considerable production of IL-10 in Mt60-immunized diabetes-resistant BiozziABH mice (I-Ag7). In addition, CBA/J mice (I-Ak) were reported to produce similar levels of IL-10 after immunization with murine or chlamydial hsp60 (40) when compared with Mt60-immunized BiozziABH or young NOD mice. These results argue against an active role of hsp60-induced IL-10 in the induction of diabetes. Moreover, these results demonstrate that the peripheral T cell repertoire of various mouse strains contains (self) hsp60-recognizing T cells, whether or not the mouse strain is susceptible to autoimmune diabetes. NOD-asp mice demonstrated a clear delay in insulitis and progression to diabetes but displayed equal kinetics on IL-10 production in time when compared with age-matched NOD-wt mice. Apparently, (self) hsp60-recognizing T cells have not been eliminated in the thymus via I-Ag7asp. In line with our results, Singer and colleagues, demonstrating that spleen cells from Abd transgenic NOD mice, recognized m60 equally well in proliferation assays as NOD-wt mice (17). Collectively, these results suggest that despite the presence of protective MHC class II molecules, (self) hsp60-recognizing T cells are positively selected via I-Ag7 and remain functional in the periphery. Interestingly, NOD-asp mice produced similar levels of IL-10 when compared with NOD-wt mice at 11 wk of age, when the majority of islets even lack cellular infiltrates. At 15 wk of age, NOD-asp mice produce higher levels of IL-10, compared with NOD-wt mice, while displaying only minor peri-islet insulitis. As a consequence of the expression of the transgene, NOD-asp mice are perhaps able to induce IL-10-producing regulatory cells more rapidly or efficiently in response to only minor inflammatory events in the pancreas. In combination with possibly enhanced negative thymic selection of potentially autoaggressive T cells via the I-Ag7asp transgene (13, 15) and cytokines produced by I-Ag7asp-restricted T cells (19), this could favor a more stable balance between autoimmunity and immune regulation in the pancreatic islets, inhibiting subsequent progression of insulitis to overt diabetes. Because we have not analyzed recognition of other autoantigens in wild-type and transgenic NOD mice, we cannot exclude the possibility that additional regulatory T cells (e.g., selected via I-Ag7asp) with specificities other than hsp60 are participating in regulation of insulitis.
Collectively, our results argue against a role of endogenous hsp60 as a primary autoantigen in spontaneous diabetes in the NOD model. More likely, (self) hsp60 T cell reactivity is a consequence of inflammatory events in the pancreatic islets irrespective of the presence of protective MHC class II molecules.
Acknowledgments
We thank Drs. R. Tisch and R. van der Zee for generously providing us the murine hsp60 expression system and recombinant Mt60 protein, respectively. A. Bloemendal and M. Boomsma are greatly acknowledged for excellent technical assistance, Dr. M. Schipper for help in statistical analysis, and J. Brandenburg for managing the NOD/LtJ/eur colony.
Footnotes
-
↵1 This work was supported by Grant 96.116 from the Diabetes Fonds Nederland; M.H.M.W. is a Fellow of the Dutch Royal Academy of Arts and Sciences.
-
↵2 Address correspondence and reprint requests to Dr. Astrid G. S. van Halteren, Department of Immunohematology and Blood Transfusion, E3Q LUMC, PO Box 9600, 2300 RC Leiden, The Netherlands. E-mail address: agsvanhalteren{at}lumc.nl
-
↵3 Abbreviations used in this paper: IDDM, insulin-dependent diabetes mellitus; hsp, heat shock protein; NOD, nonobese diabetic; wt, wild type; Mt60, Mycobacterium tuberculosis hsp60; m60, murine hsp60; HEL, hen egg lysozyme; DDA, dimethyl dioctadecyl ammonium bromide; SI, stimulation index; PBT, PBS containing 0.05% Tween 20.
- Received March 13, 2000.
- Accepted August 21, 2000.
- Copyright © 2000 by The American Association of Immunologists
References
In this issue
