|Home | About | Journals | Submit | Contact Us | Français|
Murine experimental autoimmune thyroiditis (EAT) is a model for Hashimoto’s thyroiditis, an organ-specific autoimmune disease characterized by mononuclear cell infiltration and destruction of the thyroid gland. Susceptibility to EAT is MHC-linked, and influenced by CD4+CD25+Foxp3+ regulatory T cells (Tregs). Treg depletion enables thyroiditis induction with mouse thyroglobulin (mTg) in traditionally-resistant mice and mTg-induced, Treg-mediated tolerance protects against EAT induction in genetically-susceptible mice. Here, we demonstrate the existence of naturally-existing CD4+CD25+Foxp3+ Tregs (nTregs) influencing thyroiditis development in naive susceptible mice and that induction of thyroiditis in these mice involves overcoming peripheral homeostatic immune suppression by nTregs. Additionally we demonstrate that nTregs are required for induction of antigen-specific tolerance, indicating that induced EAT tolerance is a result of activation of naturally-existing nTregs rather than de novo generation of induced Tregs (iTregs). Examination of several potential costimulatory molecules previously described as involved in peripheral activation of Tregs demonstrates a critical role indeed for CTLA-4 in the activation of nTregs leading to development of EAT tolerance and providing a mechanism for mTg-induced Treg activation during tolerance induction. Together, these data reinforce the important role of Tregs in mediating self tolerance, and illuminate a potential mechanism for their therapeutic expansion in induced tolerance.
The importance of regulatory T cells (Tregs) in suppressing autoreactive T cells in the periphery, and thus inhibiting autoimmunity, has been studied extensively in recent years. Several models have demonstrated that a naturally-existing pool of thymus-derived CD4+CD25+Foxp3+ T cells (nTregs) are essential in maintaining peripheral tolerance. nTregs inhibit multi-organ autoimmunity induced by the transfer of CD4+CD25− T cells into nude mice , and their depletion in vivo permits induction of autoimmunity, including thyroiditis, in otherwise resistant mouse strains [1–4]. The therapeutic potential of Tregs was demonstrated by the ability of transferred polyclonal nTregs to ameliorate ongoing colitis and graft-versus-host disease , furthering interest in the mechanisms of Treg activation and expansion. It is clear that the transcription factor Foxp3 is both necessary and sufficient for Treg development and function [6–8], though the signals involved in persistence and activation of nTregs following their emergence from the thymus are not entirely defined. Evidence indicates that TCR stimulation by cognate antigen and IL-2 are critical [9–12]. Additionally, there is evidence that proper costimulation is required for optimal development and peripheral maintenance of Tregs; mice deficient in B7 , as well as B7 ligands CD28 [14, 15] and CTLA-4 [16–18], or CD40-CD40L interactions [19, 20] have decreased numbers of nTregs, and exhibit increased autoreactivity.
Our approach to studying Tregs has been to use an autoimmune disease with a well-defined MHC class II-based susceptibility profile and a known autoantigen for induction of disease as well as tolerance. Experimental autoimmune thyroiditis (EAT), a mouse model of the prevalent Hashimoto’s thyroiditis, provides an excellent prototype to study peripheral function of Tregs and their role in natural and induced tolerance. In genetically-susceptible mice, mononuclear cell infiltration and subsequent destruction of the thyroid can reliably be induced by challenge with the principal thyroid antigen, thyroglobulin (Tg), and adjuvant . We have previously described the presence of nTregs inhibiting EAT induction in genetically-resistant mice , and hypothesize that nTregs mediating self tolerance also exist in genetically-susceptible mice inhibiting spontaneous autoimmunity, but are insufficient to resist challenge with antigen and adjuvant. Importantly, genetically-susceptible mice can be rendered resistant to subsequent challenge by pretreatment with exogenous soluble mouse thyroglobulin (mTg)  or endogenous mTg released by infusion of thyroid-stimulating hormone (TSH) via an osmotic pump . The antigen-specific EAT tolerance is mediated by CD4+CD25+ Tregs [24–26], and requires a critical period of 2–3 days to establish [22, 23]. This short window of tolerance induction led us to hypothesize that nTregs are activated and/or expanded by antigen stimulation along with specific co-stimulatory signals to strengthen self tolerance. We reasoned that the period for tolerance induction is too short for de novo generation of Tregs from the thymus  and concurrent administration with tolerogenic mTg of pro-inflammatory stimuli such as IL-1, anti-GITR mAb, or anti-CD137 mAb block tolerance induction [26–28].
In this report, we determined the presence of nTregs in genetically-susceptible mice and examined their role as precursors for induced tolerance to EAT and the potential mechanisms for their activation during tolerance induction. We demonstrate that these nTregs are CD4+CD25+Foxp3+ and their presence are critical for induction of EAT tolerance, supporting our hypothesis that EAT tolerance results from activation of pre-existing nTregs. We further find that blockade of CTLA-4 signaling specifically inhibits development of EAT tolerance, indicating a mechanism by which nTregs are activated during tolerance induction by endogenous or exogenous mTg, resulting in resistance to autoimmunity.
Female CBA/J (H2k) (Harlan Sprague-Dawley via C. Reeder, NIH) were used at 6–10 weeks of age. Mice were kept on acidified, chlorinated water.
Mouse thyroglobulin (mTg) was prepared from frozen thyroids fractionated on a Sephadex G-200 column as described previously . Prepared mTg was checked for the presence of LPS by Limulus amebocyte assay (a 100 µg dose of mTg contained <1 ng LPS), and diluted to working concentrations in nonpyrogenic saline. For tolerance induction, mTg was deaggregated (dmTg) prior to administration by ultracentrifugation at 100,000g at 4°C for 60 min in a Beckman SW50.1 rotor. Protein concentration of the supernatant was measured spectrophotometrically at 280nm and diluted in nonpyrogenic saline to working concentration. Tg epitope T4(2553) (STDDT4ASFSRAL), a non-immunodominant thyroiditogenic epitope shared by mTg and human Tg, was utilized to measure the in vitro proliferative response to a specific thyroiditogenic epitope .
For in vivo blocking of costimulatory molecules, CD28 (PV1.28, hamster IgG), CTLA-4 (UC10.4F10, hamster IgG2) and CD40L (MR-1, hamster IgG) mAbs were produced from hybridoma cell cultures (ATCC, Manassas, VA) in CELLMAX (Cellco, Laguna Hills, CA) or FiberCell (Fiber Cell Systems, Frederick, MD) modules and harvested according to manufacturer’s instructions. Harvested mAbs were twice purified by ammonium sulfate precipitation before i.v. administration. Hamster IgG (Fitzerald Industries, Concord, MA) was used as control.
CD25 (PC61, rat IgG1) mAb was produced for in vivo depletion of CD25+ Tregs, and purified either by 2x ammonium sulfate precipitation of supernatants from FiberCell modules, or used as ascites fluid from PC61-injected nude mice (Harlan Bioproducts for Science, Indianapolis, IN). CD25 mAb concentration was determined by ELISA with anti-ratλ (BD Pharmingen, San Jose, CA), and two 1-mg doses of CD25 mAb or control rat IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) were given i.v. 4 days apart, as previously described . CD4+CD25+Foxp3+ T cell depletion (usually 70–90% of CD4+CD25+ cells) was monitored by peripheral blood leukocytes (PBL) collected 6 days after anti-CD25 injection and analyzed by flow cytometry with slight modification of a previously described protocol . Briefly, cells were first labeled with FITC-CD4 mAb (GK1.5, rat IgG2b, eBioscience, San Diego, CA) and PE-CD25 mAb (7D4, rat IgM, Southern Biotech, Birmingham, Al), and then fixed and permeabilized per manufacturer’s instructions prior to labeling with PE-Cy5-Foxp3 mAb (FKJ-16S, rat IgG2a, eBioscience). Events were acquired uncompensated on a FACScan flow cytometer (BD, San Jose, CA) and analyzed with FlowJo software (Tree Star, Inc., Ashland, OR).
EAT tolerance was induced either with exogenous or endogenous mTg. Tolerance induction with exogenous was accomplished by administration of two 100-µg doses of dmTg administered i.v. 7 days apart, with the second dose given at least 3 days prior to challenge with EAT induction protocol to allow full establishment of tolerance . Tolerance induction with endogenously released mTg was accomplished by infusion of bovine TSH (Sigma) via a mini osmotic pump (model 2001; Alzet, Cupertino, CA) implanted i.p. under brief ether anesthesia as previously described . The pumps secreted 0.25 IU TSH/day at the rate of 1 µl/hr for 7 days. Sham operated mice were implanted with saline-filled pumps. All pumps were implanted at least 7 days prior to treatment with CD25 or control Abs.
EAT was induced using a protocol of two doses of 40 µg mTg followed 3 h later by 20 µg Salmonella enteritidis LPS given i.v. on days 0 and 7 . For mice depleted of natural Tregs, a protocol of repeated injections i.v. of 20 µg mTg, without adjuvant, on 4 consecutive days per week for a period of 4 weeks was used . Mice were sacrificed on day 28 or 35, and EAT was assessed by histologic examination of thyroids, in vitro lymphocyte proliferation to mTg, and anti-mTg production.
Thyroid specimens were sectioned vertically through both thyroid lobes (50–60 sections from 7–10 step levels), and stained with hematoxylin and eosin. The extent of mononuclear cell infiltration was assessed based upon the percentage of thyroid gland involved with lymphocytic infiltrate; a 10% or higher thyroid involvement is accompanied by destruction of thyroid architecture .
Anti-mTg levels were assessed by analysis of sera collected from tail artery and stored at −20°C until analysis. Anti-mTg Ab was detected with plate-bound mTg (1 µg/well) and alkaline phosphatase-conjugated goat anti-mouse Ig (Sigma); the enzymatic reaction was measured at 405 nm.
Splenocytes were cultured in triplicate in RPMI (RPMI 1640, supplemented with 25 mM HEPES buffer, 100 U/ml penicillin, 100 µg/ml streptomycin, 2 mM L-glutamine, 50 µM β-mercaptoethanol) + 1% normal mouse serum (NMS) in flat-bottomed 96-well plates at 6×105 cells/well, either with or without stimulation, for 5 days at 37°C, 6% CO2 . Cells were stimulated with mTg or mTg epitope T4(2553) at indicated concentrations. Cultures were pulsed with 1.0 µCi/well [3H]thymidine 18–20 h prior to harvest onto glass fiber filter paper (Tomtec Mach3Man Cell Harvester). [3H]thymidine uptake was assessed using a Microbeta Plus 1450 liquid scintillation counter (LKB Wallac). Stimulation index was calculated as mean cpm±SE with antigen divided by mean cpm without antigen.
Cell populations from naive or tolerized mice were first depleted of CD8+ cells in vivo with two 320-µg doses of CD8 mAb (YTS 167.9) and then enriched for CD4+CD25+ T cells by in vitro cell separation as described previously . Briefly, splenocytes were panned on CD11b mAb (YBM 6.1.10, rat IgM), and B cell mAb (YBM 5.10.2, rat IgM, or 187.1, rat IgG, ATCC)-coated petri dishes to remove macrophages and B cells, respectively. CD4+CD25+ cells were positively selected from the remaining cells by labeling with PE-7D4, magnetic microbeads (Stem Cell Research, Vancouver, BC), and separated according to manufacturer’s instructions. Viability and purity of CD25+ and CD25− fractions were assessed by trypan blue staining and FACS analysis (average viability >90% at 75–90% purity). Fractionated cells were co-cultured at graded doses of 3–9×104 cells with 4×105 mTg-primed splenocytes/well. Cells were cultured and proliferation was assessed as described.
Histologic data were analyzed non-parametrically using the Mann-Whitney U test. In vitro data were compared by Student’s t test. P values <0.05 were considered statistically significant.
We have previously demonstrated that susceptible mice can be made to withstand EAT induction through administration of exogenous mTg and the activated Tregs that mediate EAT tolerance are CD4+CD25+Foxp3+ [25, 26]. The strong resistance was established in 2–3 days after two (days −10 & −3), but not one (day −3), dmTg doses prior to challenge with mTg and LPS as adjuvant . On the other hand, when the endogenous mTg level was raised above baseline via an osmotic pump containing TSH, the pump can be removed after 3 days and tolerance was nevertheless established . This difference in time requirement between exogenous and endogenous mTg was shown to be due to the rapid clearance and short t½ of injected mTg, and one dose was insufficient to maintain a level higher than baseline for the required 3 days, in contrast to continuous TSH infusion for 3 days [23, 32]. This very short time interval precludes de novo induction and large expansion of Tregs. Thus, it is possible that injection of dmTg into susceptible mice expand a pool of nTregs specifically maintained by endogenous mTg.
To determine if tolerance induced by endogenous mTg release after infusing TSH was also mediated by CD4+CD25+Foxp3+ Tregs, mice were implanted i.p. with a miniosmotic pump secreting bovine TSH. After establishing tolerance, the mice were treated with anti-CD25 or control rat IgG, or left untreated. Peripheral leukocytes were monitored for Treg presence 6 days later. CD25 mAb-treated mice showed a significant reduction in CD4+CD25+ cells from peripheral lymphocytes (Fig. 1A), and analysis of Foxp3 expression showed a nearly complete reduction of CD4+CD25+Foxp3+ Tregs (Fig. 1A), as expected. Some residual CD4+Foxp3+CD25−cells remained however, as the CD25 mAb did not target those cells.
After tolerance induction and Treg depletion, all mice were immunized with mTg and LPS. As seen in Fig. 1B and C, mice implanted with control saline pumps had significantly higher thyroiditis scores and mTg Ab levels than mice implanted with TSH pumps, demonstrating that TSH infusion resulted in EAT tolerance. TSH-induced tolerance was abrogated by anti-CD25 treatment, with mononuclear cell infiltration of the thyroid and mTg Ab production similar to levels observed in control mice given saline pumps. Thus, the depleted CD4+CD25+Foxp3+ Tregs were responsible for tolerance induced by endogenous mTg release (Fig. 1B), similar to tolerance induced by exogenous soluble mTg. As mentioned above, this TSH-induced tolerance correlates with significant increases in circulating mTg levels, corroborating the specificity of exogenous mTg-induced tolerance, which cannot be induced by other autoantigens . The encounter with peripheral self antigen to maintain peripheral tolerance, and the short 3-day exposure to enhance resistance to EAT induction reinforce our hypothesis that induced tolerance results from activation or expansion of pre-existing nTregs.
Tolerance development requires a critical period of 3 days, during which we hypothesize that nTregs are activated or expanded. Previously, we were unable to observe an expansion in the number of CD4+CD25+ T cells by flow cytometry of spleen or PBL following tolerance induction via administration of soluble mTg . Presumably this is due to the relatively large number of CD4+ T cells expressing CD25 (~10%) in naive mice, which includes both Tregs to other autoantigens and normally activated T cells, compared to the small number of Tregs that would be expanded specifically by tolerogenic mTg within 3 days. Here, we use Foxp3 expression as a more specific marker of Tregs to examine the potential increase in the Treg population following tolerance induction by exogenous mTg. As illustrated by a representative experiment in Figure 2A, CD4+ cells from tolerized mice had a greater number of Foxp3+ cells (total Foxp3+ cells are quantified by adding upper and lower right quadrants of the FACS plot), compared to naive mice (representative examples of 13.2% versus 11.7%, respectively). The trend of increased Foxp3+ cells among CD4+ cells from tolerized mice was consistently observed, though variation among individual mice prevented the difference from reaching statistical significance (P=0.062).
We next investigated functional expansion of nTregs during tolerance induction by comparing the capacity of CD4+CD25+ T cells from naive and tolerized mice to suppress the proliferative response of thyroiditogenic T cells in vitro. Graded numbers of CD4+CD25+ T cell-enriched or -depleted populations from naive or tolerized mice were co-cultured with 4×105 mTg-primed splenocytes, and the in vitro proliferative response to mTg was measured. Fig. 2B illustrates a representative example of several experiments; CD4+CD25+ T cells from naive and tolerized mice demonstrated comparable in vitro suppressive capability. We also examined the ability of CD4+CD25+ T cells from naive and tolerized mice to suppress the proliferative response of primed T cells to a pathogenic T cell epitope T4(2553) , should a more subtle difference in suppressive function be revealed by an individual epitope, rather than multiple epitopes on mTg . However, no differences were observed in the capacity of CD4+CD25+ T cells from tolerized mice to suppress proliferative responses of thyroiditogenic T cells to T4(2553), compared to CD4+CD25+ T cells from naive mice (Fig. 2C).
Our hypothesis of Treg-mediated EAT tolerance development suggests that induced tolerance results from the activation of mTg-specific nTregs. While we have demonstrated Treg inhibiting thyroiditis development in thyroiditis-resistant mice , it was necessary to demonstrate the presence of nTregs influencing susceptibility to thyroiditis induction in naive EAT-susceptible CBA mice. Because of their natural susceptibility, challenge of naive CBA mice with mTg and adjuvant generally results in 100% incidence of thyroiditis , necessitating the use of a less potent induction protocol. Repeated doses of mTg in the absence of adjuvant typically results in low levels of mononuclear cell infiltration of the thyroid in approximately 50% of EAT-susceptible mice , and could serve as a more sensitive assay for the in vivo influence of naturally-existing Tregs on thyroiditis induction.
CD4+CD25+ T cells were depleted as described previously  by i.v. administration of two 1-mg doses of CD25 mAb on days −14 and −10 prior to challenge. Treg depletion was verified by FACS analysis of PBL on day −4. Treatment of mice with CD25 mAb resulted in a 70–90% reduction in CD4+CD25+ T cells in the peripheral blood or spleen (data not shown). Following CD25 mAb treatment, 20 µg mTg was administered i.v. daily 4×/wk for 4 wk (days 0, 1, 2, 3, 7, 8, 9, 10, 14, 15, 16, 17, 21, 22, 23 and 24), and thyroiditis development was assessed by examination of mononuclear cell infiltration of the thyroid, production of anti-mTg antibody, and in vitro response of splenocytes to antigen on day 35. The incidence and severity of thyroiditis induced by mTg administration without adjuvant were both increased by depleting CD4+CD25+ T cells; mononuclear cell infiltration was observed in 16/24 (67%) mice and destruction of thyroid architecture in 14/24 (58%) mice treated with CD25 mAb, compared to 11/24 (46%) and 6/24 (25%), respectively, of controls (Fig. 3A) (P < 0.02). Follicular disruption was minimal in control mice, involving 15% or less of the thyroid gland (Fig. 3C), but was increased in CD25 mAb-treated mice, with up to 30% involvement of the gland (Fig. 3D). Increased susceptibility to thyroiditis in CD4+CD25+ T cell-depleted mice was also demonstrated by enhanced mTg Ab level (Fig. 3E), and greater in vitro proliferative response to mTg and T cell epitope T4(2553)  (Fig. 3F). These data demonstrate the presence of nTregs influencing susceptibility of thyroiditis induction in EAT-susceptible CBA mice.
The presence of CD4+CD25+ nTregs inhibiting the induction of EAT in CBA mice (Fig. 3) provides the first requisite of our hypothesis that induced EAT tolerance is the result of strengthening the naturally-existing tolerance to mTg rather than the induction of a novel mechanism of suppressing autoimmunity. Since we could not directly observe an expansion in the Treg compartment following tolerance induction (Fig. 2), we further tested this hypothesis by depleting nTregs prior to injecting tolerogenic mTg to directly assess their role. CBA mice were depleted of CD4+CD25+ T cells by two 1-mg doses of CD25 mAb given i.v. on days −24 and −20, allowing for depletion and monitoring of CD4+CD25+ nTregs and clearance of CD25 mAb prior to tolerizing doses of 100 µg dmTg on days −10 and −3. Mice were subsequently challenged with 40 µg mTg and 20 µg LPS on days 0 and 7, and thyroiditis severity was assessed on day 28. Depletion of CD4+CD25+ cells prior to dmTg administration resulted in an inability of 10/11 (91%) mice to become tolerized and resist EAT induction, while only 1/12 (8%) rat IgG-treated mice failed to establish tolerance (P < 0.01) (Fig. 4). The inability of CD4+CD25+ T cell-depleted mice to respond to EAT tolerance induction supports our early hypothesis that EAT tolerance results from an activation or expansion of pre-existing Tregs by encounter with self antigen in the periphery .
Our hypothesis of EAT tolerance resulting from an activation or expansion of naturally-existing Tregs also implies that a particular mechanism should be responsible for tolerogenic activation of Tregs upon encounter with self antigen. Roles for CD28, CD40L, and CTLA-4 in the peripheral maintenance of Tregs have been described in other models of Treg function (see Introduction), and we examined the effect of blockade of these signaling pathways on the activation of nTregs during EAT tolerance induction. A protocol of six doses of mAbs, 250 µg/dose of CD28 mAb and CD40L mAb, based upon previous observation of their in vivo activity , and 1 mg/dose of CTLA-4 mAb, based upon initial titration experiments (data not shown), given i.v. on days −16, −15, −14, −9, −8, −7, was used to surround the window of tolerance induction by 100 µg of dmTg on days −15 and −8. Mice were challenged with mTg and LPS on days 0 and 7 and thyroiditis severity was assessed on day 28. As illustrated in Fig. 5, in vivo administration of CD28 mAb or CD40L mAb had no effect on tolerance induction as compared to control hamster IgG; 0/12 mice in the CD28 mAb-treated group, and only 1/12 mice in the CD40L mAb-treated group had any mononuclear infiltration of the thyroid. However, administration of CTLA-4 mAb at the time of tolerization resulted in low levels of thyroid infiltration in 7/12 (58%) mice, compared to 0/10 control hamster IgG-treated tolerized mice (P < 0.02). The specific inhibition of tolerance induction by CTLA-4 mAb indicates that signaling through CTLA-4 is an important mechanism for tolerogenic activation or expansion of CD4+CD25+Foxp3+Tregs in vivo.
The role of CTLA-4 in mediating suppression of already primed thyroiditogenic T cells by CD4+CD25+ Tregs is difficult to evaluate, as activated thyroiditogenic T cells would be expected to express CTLA-4 on the cell surface, and mAb would act upon both Tregs and thyroiditogenic T cells. We attempted to observe an effect of CTLA-4 mAb on the suppressive capabilities of Treg by in vitro co-culture of 1×105 CD4+CD25+ T cells enriched from tolerized mice with 4×105 mTg-primed splenocytes; addition of 50 µg/ml or 100 µg/ml CTLA-4 mAb had no effect on the ability of CD4+CD25+ Tregs to suppress the in vitro response of mTg-primed T cells (data not shown). While we were unable to discern a role for CTLA-4 in mediation of Treg function in tolerance, we demonstrate that CTLA-4 is important for proper co-stimulation of nTreg during tolerance induction, indicating a potential mechanism for the functional expansion of Treg responsible for induced tolerance to EAT.
Our previous investigations of Tregs have focused on their mediation of induced tolerance inhibiting subsequent induction of thyroid-specific autoimmunity. Development of CD4+ Treg-mediated resistance to EAT induction requires a period of 2–3 days, either following the administration of exogenous mTg [22, 24–26], or the physiologic release of endogenous antigen by infusion of TSH [23, 32]. This critical period of antigen-specific tolerance induction correlates directly with increased circulating mTg levels and is mediated by CD4+ Tregs . These findings supported our early hypothesis that, during this short window of tolerance induction, either by exogenous or endogenous antigen, Tregs mediating tolerance arise either by expansion and/or differentiation of pre-existing nTregs, and that proper activation of nTregs required interaction with self antigen as well as proper costimulation providing a tolerogenic context for Treg activation [26, 28]. We showed also that tolerance induced with exogenous mTg was mediated by CD4+CD25+Foxp3+ Tregs. Here, we demonstrate that CD4+CD25+Foxp3+ Treg-mediated tolerance likewise develops as a result of encounter with endogenous self antigen (Fig. 1).
To examine what happened to the Treg population after exposure to tolerogenic mTg, we attempted to assay for expansion of CD4+CD25+ Tregs both by phenotypic analysis of peripheral T cells as well as by functional analysis. We examined the numbers of peripheral Tregs more specifically by focusing on expression of Foxp3, a marker specific for Tregs [6, 7]. Rather than molecular analysis by RT-PCR, we analyzed by three-color flow cytometry in order to evaluate Foxp3 expression concurrently with other Treg markers to avoid problems of impurities from cell separation, and to gauge the level of fluorescence intensity. A consistent trend of increased numbers of CD4+CD25+Foxp3+ T cells was observed in tolerized mice (Fig. 2A), but it did not reach statistical significance. It is likely that the expansion of a low number of mTg-specific Tregs during the 2–3 days of tolerance induction is too small to be accurately measured within biologic variability among animals; our hypothesis, along with our previous data demonstrating the antigen-specific nature of induced tolerance [22, 32], suggests that only mTg-specific Tregs would be expanded. Functional evaluation of Tregs from tolerized mice also failed to demonstrate an expansion of Tregs; titration by number up to 3-fold did not reveal any difference in their suppressive function (Fig. 2B–C). The lack of a notable difference in suppressive function in vitro may be explained in two ways. First, nTregs from naive mice could be activated in vitro, potentially by mTg in culture and IL-2 produced by the activated thyroiditogenic CD4+CD25− T cells [34, 35]. In vivo tolerance to thyroiditis is antigen-specific, as mice given equivalent doses of liver extract demonstrate no tolerance to thyroiditis induction, and homologous human Tg induces tolerance only to determinants shared with, but not unique to, mTg , unlike mice treated with mTg  . In vitro assays appear to be different, however, as moderate non-specific suppression of ovalbumin-primed splenocytes by Tregs from mTg-tolerized mice can be observed (data not shown). Second, our assay for in vitro suppressive function may not adequately model the in vivo changes in Tregs which suppress thyroiditis development. Treg migration, localization in the thyroid and lymphoid tissues, and development in local microenvironment are all components of in vivo Treg function that are not adequately represented by in vitro co-culture. These parameters are likely altered by Treg activation during tolerance induction, and warrant further investigation.
A requisite of our hypothesis on tolerance development is the presence of nTregs inhibiting thyroiditis in mice prior to tolerance induction. We have previously demonstrated nTregs inhibiting thyroiditis induction in genetically-resistant mice . To reveal thyroiditis-inhibiting nTregs in EAT-susceptible mice, we utilized an adjuvant-free model of thyroiditis induction . This model induces less severe disease in 50% of mice compared to 100% incidence with more severe thyroiditis after immunization with mTg and LPS, enabling the evaluation of less pronounced changes in susceptibility. Indeed, in vivo administration of CD25 mAb prior to mTg immunization lowered the threshold for thyroiditis induction with an adjuvant-free EAT induction protocol (Fig. 3), demonstrating the presence of nTregs normally inhibiting thyroid autoimmunity.
Our hypothesis that rapid EAT tolerance development is predicated on the preexistence of nTregs is further strengthened by the requirement of nTreg presence at the time of EAT tolerance induction. Indeed, in vivo depletion by mAb administration eliminates the ability to establish resistance to EAT (Fig. 4). Because of the strong increase in their suppressive ability in vivo, we have previously considered both quantitative and qualitative changes as likely important in the development of induced EAT tolerance; nTreg activation may include both expansion and differentiation. While our data directly demonstrates the necessity for nTregs for tolerance induction, it does not eliminate the possibility of peripheral conversion of non-Tregs into Tregs during tolerance induction. It is possible that nTregs are both expanded as well as participate in conversion of peripheral T cells into Tregs. Recent experimental evidence suggests that Tregs mediating tolerance to thyroiditis induction are induced by presentation of mTg by semi-mature dendritic cells [37, 38], and Tregs mediate their function in vivo, as least in part, through interaction with dendritic cells resulting in down-regulation of CD80 and CD86 [39, 40]. These data suggest that the role nTregs play in tolerance induction may be through providing a tolerogenic context for antigen presentation, which is compatible with our data indicating a critical role for nTregs in these processes. The degree of tolerance induction dependent on direct expansion of nTregs compared to that dependent on peripheral conversion requires further study.
The possibility has been put forth that functional inactivation, rather than numerical depletion, may occur with administration of CD25 mAb . However, in our hands, our previous studies showed that CD25 mAb treatment resulted in depletion of CD4+CD25+ Tregs in PBL as assayed by flow cytometry with a mAb to a different CD25 epitope . Moreover, we showed here the simultaneous reduction of Foxp3+ Tregs from the CD25+ population (Fig. 2A), demonstrating that anti-CD25 treatment indeed depletes Tregs. This possibility notwithstanding, the elimination of functional Tregs dramatically interferes with self tolerance, either natural or induced. Interestingly, a population of Foxp3+CD25− cells remained after CD25 mAb treatment (Fig. 3), though this population appeared insufficient to prevent thyroiditis induction in any experiment. It would be of interest to further characterize the role of these cells in maintaining peripheral tolerance.
The exact mechanism by which nTregs are triggered to activate or expand during tolerance induction is unclear. Thyroiditis susceptibility has long been ascribed to the H2A allele, presumably through proper antigen presentation for the generation of autoreactive T cells in the thymus and their activation in the periphery. We recently observed the extension of MHC class II restriction to Tregs in a transgenic H2E+ H2A–Knockout mouse , indicating that the influence of MHC class II genes on susceptibility to autoimmunity is not limited to the generation and activation of autoreactive T cells, but also includes the same restriction on nTregs. In addition to the requirement of MHC class II and TCR engagement for thymic generation, we consider likely that peripheral triggering of nTregs also depends upon TCR engagement by specific antigen presented in the context of proper MHC. The recently reported requirement of islet antigen for Treg activity in a Type I diabetes model of NOD mice is in agreement with our interpretation . Induction of EAT tolerance is antigen-specific [22, 23], in agreement with recent reports suggesting mTg presentation in a tolerogenic context by semi-mature dendritic cells [37, 38], which, with appropriate costimulatory signals, can induce specific tolerance. B7-CTLA-4 interactions generated CD4+CD25+ Tregs capable of inhibiting transplant rejection  and colitis , and targeted delivery of CTLA-4 mAb to the thyroid inhibited thyroiditis development, possibly through the generation of Tregs [43, 44]. Indeed, our studies have shown that administration of proinflammatory substances, such as polyadenylic-polyuridic acid complex (poly A:U) , IL-1 , or IL-12 , or signaling through specific costimulatory molecules CD137  and GITR , inhibits EAT tolerance induction. Here, we demonstrate that administration of mAbs to CD28 or CD40L, at doses previously used to interfere with the activation of thyroiditogenic T cells , has no effect on the ability to induce tolerance. In contrast, CTLA-4 mAb specifically interferes with the activation of naturally-existing Tregs by dmTg administration (Fig. 5). The inability of CTLA-4 mAb to completely inhibit tolerance induction in all mice may be related to problems with in vivo administration affecting bioavailability of the mAb; all mice that received the hamster CTLA-4 mAb in vivo developed inflammation and desquamation at the injection site which resolved within 7 days. This effect was not observed with other hamster mAbs, or control hamster IgG, and precluded the use of higher doses of the mAb. Also, the possibility of compensatory or additional signaling pathways cannot be discounted. In particular, several models have implicated the participation of TGF-β in Treg activation [45, 46], a possibility that requires further investigation in EAT tolerance.
In addition to Treg activation, CTLA-4 has also been implicated in the suppressive function of Tregs; CTLA-4−/− mice develop a severe multiorgan autoimmunity [47, 48] attributable to the function of CTLA-4 on Tregs . Additionally, CTLA-4 mAb interferes with Treg-mediated suppression of autoimmunity in vivo and in vitro [49–52]. In recent clinical trials, high and repeated doses of CTLA-4 mAb led to manifestations of a variety of autoimmune symptoms [53, 54], illustrating the marked influence of CTLA-4 on autoimmune suppression. The recently observed role of CTLA-4 on Tregs decreasing the stimulatory capability of DCs by causing down-regulation of costimulatory molecules [39, 40], has particular implications in our model of tolerance induction; this connects our observations of the critical role of nTregs during tolerance induction to generate the proper tolerogenic context for the presentation of autoantigen. However, it is still not clear whether downregulation resulting in functional Treg expansion is by expansion of pre-existing mTg-specific Tregs, and/or by peripheral conversion of non-Tregs.
The potential confounding variable of the effect of CTLA-4 mAb administration on thyroiditogenic T cells prohibited our use of the mAb to assess the in vivo effect of CTLA-4 blockade during the mediation of established tolerance. However, contrary to reports in other models, we did not observe any interference with the in vitro suppressive function of Tregs in the presence of CTLA-4 mAb at concentrations up to 100 µg/ml. As mentioned above, our co-culture assay might not adequately assess the mechanism of Treg function in vivo. Additionally, CTLA-4 mAb might not adequately inhibit signaling through CTLA-4, redundant mechanisms might exist to compensate for loss of CTLA-4 signaling, or Tregs from tolerized mice had already been well activated in vivo, and further CTLA-4 signaling to exert their suppressive effect was not required.
Whatever the mechanisms responsible for triggering Treg activation and expansion during tolerance induction, it is clear that an increase in their suppressive ability occurs in vivo, as Treg function becomes greatly augmented to resist thyroiditis induction by mTg and adjuvant. The data support our long-standing working hypothesis that induced EAT tolerance is the result of an activation of pre-existing nTregs . These CD4+CD25+ nTregs serve to control EAT development in both susceptible and resistant strains [4, 26], although they do not supersede MHC class II restriction . The activation of nTregs that occurs during EAT tolerance induction requires signaling through CTLA-4 for optimal tolerance development, indicating the importance of proper costimulatory signals for nTreg activation, and underscoring the importance of the context of tolerogenic antigen for the induction of immunologic tolerance. Understanding the particular requirements for activation of nTregs and the subsequent induction of immunologic tolerance, rather than activation of autoreactive T cells and autoimmunity, is critical for rational development of new immunotherapies.
The authors thank Ms. A. M. Mazurco and Ms. Renee Wilder for preparation of histology sections and Dr. A. A. Giraldo and Daniel P. Snower for assistance in evaluating thyroid pathology. We also appreciate the kind gifts from Dr. C. Jeffries for S. enteritidis LPS, Dr. Gerald Nabozny for CD40L mAb, and Dr. H. Waldmann for the CD4, CD8, CD11b and B cell mAbs. This study was supported by grant NIDDK 45960 (Y. M. Kong).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.