Abstract
Autoimmunity often spreads in a predefined pattern during the progression of T cell-mediated autoimmune diseases. This progression has been well described in animal models and in man, but the basis for this phenomenon is little understood. To gain insight into the factors that determine this spreading hierarchy, we characterized the binding affinity of a panel of β cell-autoantigenic peptides to I-Ag7, as well as the precursor frequency, functional avidity, and phenotype of the T cells that recognize these peptides in type 1 diabetes-prone nonobese diabetic mice. We observed that autoimmunity gradually spreads from a β cell determinant, which had the largest precursor pool of high avidity T cells, to β cell determinants with progressively smaller and lower avidity T cell precursor pools. This correlation between the sequential development of spontaneous T cell autoimmunity and the frequency and avidity of autoantigen-reactive T cells suggests that the extent to which T cells were negatively selected by the self-determinants is the key factor determining the spreading hierarchy.
It has been widely observed that during the development of T cell-mediated autoimmune diseases, autoreactive T cell responses gradually spread from a limited number of determinants to many additional target tissue Ags (1, 2, 3, 4, 5). This spreading of T cell autoimmunity often follows a defined pattern during disease progression. For example, T cell autoimmunity primed by a single encephalitogenic peptide of proteolipid protein spreads intramolecularly (within proteolipid protein) and intermolecularly (to other myelin Ags) in a defined chronological order (1, 2, 3). Notably, the progression of the resulting experimental allergic encephalomyelitis (EAE)3 appears to depend on determinant spreading because relapses can be prevented by inducing tolerance to the second-wave target determinants (6). Such sequential spreading of T cell autoreactivity to myelin Ags has also been observed in patients with multiple sclerosis during disease progression (3). Therefore, understanding why certain self-determinants become immunogenic before other self-determinants, creating a “spreading hierarchy”, may provide insights into the pathogenesis of organ-specific autoimmune diseases, and assist in the design of stage-specific therapies.
Theoretically, the spreading hierarchy of spontaneous autoimmunity should reflect the different immunogenicities of self-determinants in the target tissue. Studies of immune responses to foreign determinants have shown that foreign determinants often have large pools of high avidity precursor T cells (as they did not contribute to T cell selection) and that the primary factor defining their immunogenicity is the efficiency with which they are displayed on MHC (7). Consequently, the most immunogenic foreign determinants are often those with the highest affinity for MHC. However, these rules may not apply to self-Ags, as the same factors that cause a foreign determinant to be immunogenic promote the induction of T cell tolerance to self-determinants. Indeed, the repertoire of potentially autoreactive T cells should bear the imprint, in terms of both their precursor frequency and avidity, of selection by self-determinants. Accordingly, the self-reactive T cell pool should be trimmed to leave only those T cells that interact with self-Ags below the T cells’ activation thresholds. Given this trimming of the self-reactive T cell repertoire, it is presently unclear what properties of the remaining self-reactive T cells, and of their cognate self-determinants, underlie the predefined patterns in which T cell autoreactivity spreads during the pathogenesis of organ-specific autoimmune diseases.
The nonobese diabetic (NOD) mouse spontaneously develops autoreactive T cell responses to β cell Ags at ∼4 wk of age, concurrent with the onset of insulitis (4, 8, 9, 10, 11, 12, 13). Initially, T cell autoreactivity appears to be limited in its specificity, but it gradually spreads among Ag determinants in a defined chronological pattern, often leading, several months later, to insulin-dependent diabetes mellitus (IDDM) (4, 8, 9). This sequential spreading of T cell autoreactivity among β cell Ags provides a model system to dissect the factors that determine the spreading hierarchy. Toward this, we identified a panel of different β cell Ag determinants that become involved in the autoimmune cascade at different stages of the disease process in NOD mice. We then systematically characterized the factors underlying their immunogenicity, namely, their binding affinity for MHC, as well as the precursor frequency and functional avidity of reactive T cells in preautoimmune NOD mice.
Materials and Methods
Mice
NOD mice (Taconic Farms, Germantown, NY) were bred under specific pathogen-free conditions. In our colony, ≈85% female NOD mice spontaneously develop IDDM by 30 wk of age. Only female mice were used in these studies.
Antigens
The 65-kDa form of mouse glutamic acid decarboxylase (GAD) and control Escherichia coli β-galactosidase were purified as previously described (8). The autoantigenic and immunodominant GAD peptides that were tested included GAD524–543 (also termed GADp35; SRLSKVAPVIKARMMEYGTT), GAD78–97 (also termed GADp6; KPCSCSKVDVNYAFLHATDL), and GAD217–236 (also termed GADp15; EYVTLKKMREIIGWPGGSGD) (4, 8, 14). Peptides from other key β cell autoantigens included the immunodominant peptide of the 65-kDa heat shock protein (HSP), termed HSPp277 (VLGGGCALLRCIPALDSLTPANED; Ref. 15) and the immunodominant determinant of insulin, insulin9–23 (SHLVEALYLVCGERG; Ref. 16). An immunogenic hen egg lysozyme (HEL) peptide HEL11–25 (AMKRHGLDNYRGYSL; Ref. 17) was used as a control foreign peptide and a mouse serum albumin (MSA) peptide MSA560–574 (KPKATAEQLKTVMDD; Ref. 18), which contains a nontargeted determinant, was used as a control self-Ag. All peptides were synthesized by Multiple Peptide Systems (San Diego, CA) at ≥95% purity.
Purification of the I-Ag7 molecule and peptide binding assay
The I-Ag7 molecule was affinity purified from detergent lysates of 4G4.7 B cell hybridoma by sequential negative selection with 34.1.4 and 14.4.4S mAbs followed by elution from OX-6 mAb coupled to protein A-Sepharose, as previously described (19). Briefly, the lysates of 1011 4G4.7 cells were passed on the 34.1.4 protein A-Sepharose column followed by a 14.4.4S protein A-Sepharose column, and finally loaded on to the OX-6 protein A-Sepharose column. Following extensive washing, the I-Ag7 molecule was eluted and immediately neutralized. The purity of the eluted proteins was >95% as demonstrated by two bands with m.w. ∼33,000 and ∼28,000 resolved in SDS-PAGE (corresponding to the α- and β-chains, respectively, of MHC class II molecules).
Peptides for the analysis of affinity were dissolved at 10 mM in DMSO and diluted with 25% DMSO in PBS for the assay. The peptide HEL10–23 was synthesized with two spacer residues and a biotin molecule at the NH2 terminus and used as the indicator. The indicator (500 nM), together with different concentrations (50 μM-50 pM at 10-fold dilution) of the tested peptides, was coincubated with 200 nM of I-Ag7 molecules in U-bottom polypropylene 96-well plates in binding buffer at room temperature for 48 h. The concentration of the complex of biotinylated indicator peptide HEL10–23/I-Ag7 was determined by an ELISA using prebound OX-6 Abs (10 μg/ml) and streptavidin-alkaline phosphatase and p-nitrophenylphosphate. Competition curves were plotted, and the peptide affinity for MHC molecules was expressed as the peptide concentration required to inhibit the binding of biotinylated peptide by 50% (IC50).
Immunizations
To determine the frequency of autoantigen determinant-reactive precursor T cells and the functional T cell avidity, we immunized 2-wk-old female NOD mice with a control or autoantigen peptide (50 nanomoles per mouse) in 50% CFA (Life Technologies, Grand Island, NY) in the footpad. Nine days after immunization, single cell suspensions were prepared from the draining lymph nodes, and Ag-specific T cell responses were characterized by proliferation or enzyme-linked immunospot (ELISPOT) assays.
Proliferation assay
Nine days following immunization, single cell suspensions were prepared from the draining lymph nodes. Mononuclear cells (4 × 105 cell/well) were stimulated with different concentrations (0.2–70 μM) of Ag peptides in triplicate in FCS-free HL-1 medium (BioWhittaker, Walkersville, MD) and incubated in 96-well microtiter plates at 37°C with 5% CO2 for 96 h. Medium alone (without any Ag) was used as the negative control, and purified protein derivative (PPD; 10 μg/ml) or anti-CD3 (1 μg/ml) was used as the positive control for each mouse. During the last 12–16 h of the 96-h culture period, 1 μCi [3H]thymidine was added to each well. Incorporation of label was measured by liquid scintillation counting. Lymph node T cell proliferation to a specific Ag was plotted as dose-dependent curves, and the functional T cell avidity was presented as the percentage of proliferation.
ELISPOT analysis
The frequency of Ag-specific splenic T cells secreting IFN-γ, IL-4, and IL-5 was determined using a modified ELISPOT technique as previously described (20, 21). Briefly, 106 splenic mononuclear cells per well were added (in duplicate) to an ELISPOT plate that had been coated with cytokine capture Abs and incubated with peptide (20 μM) or whole protein (100 μg) at 24 h for IFN-γ, or 40 h for IL-4 and IL-5 detection. After washing, biotinylated detection Abs were added, and the plates were incubated at 4°C overnight. Bound secondary Abs were visualized using HRP-streptavidin (Dako, Carpinteria, CA) and 3-amino-9-ethylcarbazole. Abs R4-6A2/XMG 1.2-biotin, 11B11/BVD6-24G2-biotin, and TRFK5/TRFK4-biotin (PharMingen, San Diego, CA) were used for capture and detection of IFN-γ, IL-4, and IL-5, respectively.
To characterize lymph node T cell responses in Ag-primed mice, draining lymph node mononuclear cells (4 × 105 cells per well) were challenged with different concentrations of control or autoantigen peptides, 20 μg/ml PPD, or 1 μg/ml anti-CD3. The frequency of Ag-induced spot-forming colonies (SFC) is expressed as dose-dependent curves. The proportion of high, intermediate, and low avidity T cells in an Ag-specific repertoire was determined based on the average number of SFC in response to 0.2, 0.7–2.0, and ≥2.0 μM Ag (respectively) divided by the average maximal number of Ag-specific SFC responses.
Results
Sequential development of spontaneous T cell immunity to autoantigen determinants
Previous studies of the determinant spreading of autoimmunity during the development of IDDM in NOD mice used T cell proliferation assays (8, 9), which are limited in sensitivity and may not have detected low frequency autoreactive T cells. To obtain a higher resolution analysis of the spreading hierarchy, we used an ELISPOT assay capable of detecting a single autoreactive T cell within a million, and which unlike proliferation assays, is not affected by the presence of regulatory T cells (22).
At different stages of NOD mouse development, we characterized splenic T cell responses to a panel of β cell-autoantigenic determinants. The tested β cell autoantigens included GAD78–97, GAD217–236, GAD524–543, insulin9–23, and HSPp277, which have been implicated as being key target determinants (4, 8, 14, 16, 23, 24, 25). Control Ags included an I-Ag7 binding nontargeted self-peptide from mouse serum albumin, MSA560–574, as well as a foreign peptide from hen egg lysozyme, HEL11–25.
We did not detect a splenic T cell response to any of these Ags in 3-wk-old NOD mice (Fig. 1⇓). Beginning at 4 wk of age, IFN-γ-secreting T cells reactive to whole GAD and GAD524–543 were detected. At 6–8 wk of age, spontaneous autoimmunity spread to GAD78–97 and HSPp277. By 12 wk of age, T cell responses to GAD217–236 and insulin9–23 appeared (Fig. 1⇓). All of these autoreactivities were Th1 type, with only IFN-γ SFC, and no detectable IL-4 and IL-5 SFC (data not shown). No responses to self-Ag MSA560–574 or foreign determinant HEL11–25 were detected at any age. Thus, ELISPOT analysis of T cell frequency and phenotype confirm previous observations (4, 8) that the hierarchy of determinant spreading among this panel of autoantigenic peptides is GAD524–543 (an early target), followed by GAD78–97 and HSPp277 (intermediate targets), and eventually by GAD217–236 and insulin9–23 (late targets).
Splenic T cell responses to β cell Ags develop spontaneously in a defined chronological order. Splenic T cells from 2- to 3-, 4-, 6- to 8-, and 12-wk-old female NOD mice were tested for their responses to a panel of β cell autoantigens and control nontarget tissue Ags using an ELISPOT assay. Data shown are the mean number of IFN-γ-secreting SFC per 106 splenic T cells ± SD. NOD and control BALB/c mice were tested simultaneously (in duplicate) in at least two separate experiments (n ≥ 8 total for each group). No responses were detected at any age to MSA560–574 or HEL11–25. Responses by splenic T cells from age-matched BALB/c mice to all tested Ags were at background levels (data not shown).
Affinity of autoantigen peptides for I-Ag7
To dissect the factors that determine the spreading hierarchy, we first tested whether the strength of MHC binding could determine the immunogenicity of these β cell Ag determinants. Using a competitive peptide binding assay, we observed that autoantigen peptides displayed a range of binding affinities for I-Ag7 (Fig. 2⇓). The hierarchy of control and β cell autoantigen peptide binding affinities for I-Ag7 (expressed as IC50) is MSA560–574 (0.4 μM) < insulin9–23 (0.6 μM) < HEL11–23 and GAD217–236 (0.7 μM) < GAD524–543 (0.8 μM) < GAD78–97 (1.3 μM) < HSPp277 (10.7 μM). This ranking of binding affinity of the autoantigen peptides for I-Ag7 displays no relationship to the chronological pattern in which spontaneous autoreactive T cell responses spread, either intramolecularly (within GAD) or intermolecularly (Fig. 1⇑). For example, both early (GAD524–543) and late (GAD217–236) target determinants of GAD have fairly high affinities for I-Ag7, and a late target determinant (insulin9–23), has a higher affinity for I-Ag7 than an intermediate target determinant (HSPp277). Thus, the sequential spreading of autoimmunity among β cell Ag determinants is independent of the affinity for I-Ag7 of the determinant.
Affinity of β cell autoantigen peptides binding to I-Ag7. A, Competition between biotinylated HEL10–23 and unlabeled peptides for binding to purified I-Ag7. A representative experiment of three performed is shown. B, The affinity of the indicated peptides for I-Ag7 was expressed as the concentration that inhibited binding of the indicator peptide by 50% (IC50). Bars represent the mean ± SE from three independent experiments.
The size of autoantigen-reactive precursor T cell pool correlates with the spreading hierarchy
We next tested whether the clonal size of autoantigen-specific T cells in the preimmune repertoire could explain the observed spreading hierarchy. We immunized NOD mice with a high dose of a peptide (50 nM in CFA) to maximally load the I-Ag7 molecules so that Ag presentation should not be a limiting factor. NOD mice were immunized at ≈14 days of age, before the onset of the endogenous autoimmune response. Nine days after the immunization, the frequency of peptide-specific T cells was determined in the draining lymph node by ELISPOT. Under these conditions, the induced T cell response provides a measurement of the size of an Ag-specific T cell pool in preautoimmune NOD mice.
Immunization with the control self-peptide MSA560–574 primed the lowest number of T cells, as expected for a well presented self-Ag (Fig. 3⇓). In contrast, GAD524–543 primed an even greater number of peptide-reactive T cells than the control foreign Ag determinant HEL11–25. GAD524–543 also induced a significantly greater number of T cell responses than did GAD78–97, which in turn induced more T cells than GAD217–236. Therefore, when Ag presentation is not limiting, the immunogenicity of determinants within a single autoantigen (GAD) follows the order in which autoimmunity spreads intramolecularly among these determinants during disease development. Similarly, HSPp277, an intermediate target along with GAD78–97, primed an intermediate level of T cell responses. However, the frequency of T cell responses to the late target insulin9–23 was significantly higher than that of both HSPp277 and GAD78–97. These data suggest that the size of the precursor T cell pool may be an important factor in establishing the spreading hierarchy. However, the lack of complete correlation between the size of precursor T cell pools and the spreading hierarchy suggests that some other factor(s) contribute to the establishment of the spreading hierarchy.
Frequency of precursor T cells responding to β cell autoantigen determinants. Data shown are the mean number of IFN-γ-secreting SFC per 4 × 105 lymph node cells ± SD. Mice were tested simultaneously (in triplicate) in at least two separate experiments (n ≥ 8 total for each group). All Ag-primed mice displayed similar responses in magnitude to control PPD or anti-CD3 stimulation. Ag-primed mice developed unipolar Th1 cell responses only to the injected Ag but not to any other tested autoantigen. Lymph node cells from age-matched unprimed NOD mice showed no responses to any of the Ags tested.
The frequency of T cells with high avidity matches the sequence of spontaneous T cell immunity to autoantigen determinants
During the development of self-tolerance, self-determinants shape the size of precursor T cell pools and limit the avidity of self-reactive T cells remaining in the repertoire. Notably, the immunogenicity of a determinant depends on both the clonal size and the avidity of reactive T cells. Therefore, we next examined the functional avidity of autoantigen-reactive T cells in the preimmune repertoire. High avidity T cells become activated at low peptide concentrations, and vice versa. Accordingly, functional T cell avidity for a determinant can be measured by the dose response to the peptide, and defined as the peptide dose at which a 50% maximal response is observed (26).
As described above, ≈14-day-old NOD mice were immunized with a control or autoantigen peptide in CFA, and their draining lymph node T cells were tested for proliferative responses to different concentrations of the Ag (Fig. 4⇓A). The relative strength of T cell responses primed by different concentrations of each peptide is presented in Fig. 4⇓B. The dose-response curve for GAD524–543 was similar to that of the foreign determinant HEL11–25, and both of these responses were severalfold higher than that primed by other self-determinants (Fig. 4⇓A). Both GAD524–543 and HEL11–25 induced 50% maximal T cell proliferation at a concentration of ≈1 μM. The intermediate target determinants GAD78–97 and HSPp277 required ≈8 and 13 μM (respectively), whereas the late target determinants GAD217–236 and insulin9–23 required ≈17 and 19 μM (respectively) to induce 50% of maximal T cell proliferation. Accordingly, the order of functional T cell avidity, as determined by proliferative responses, matched the developmental sequence of spontaneous T cell immunity to the tested autoantigen determinants.
A, Functional T cell avidity for β cell autoantigen determinants. The data are presented as the average cpm of 6–8 mice per group. The background level (medium alone) ranged from 1500 to 3000 cpm. The variation between wells was <15%. PPD (10 μg/ml) and anti-CD3 (0.5 μg/ml) induced a similar level of T cell proliferation (45,000–55,000 and 23,000–27,000 cpm, respectively) in every group. B, Response to different concentrations of each Ag plotted as the percentage of the maximum response to that Ag. Functional T cell avidity was determined as the peptide concentration that evoked 50% of the maximal response.
However, proliferative responses may be influenced by the presence of regulatory factors and are influenced by IL-2-driven bystander blastogenesis. Therefore, we also used dose-dependent ELISPOT assays to determine the overall functional avidity, as well as the relative frequency of high, medium, and low avidity T cells in autoantigen-reactive precursor T cell pools.
The minimal amount of Ag required to induce significant T cell IFN-γ-secreting SFC varied widely among the different tested Ags (Fig. 5⇓A). Only 0.2-μM peptide was sufficient to elicit frequent IFN-γ-secreting colonies to GAD524–543 and foreign Ag HEL11–25.In contrast, 0.7- to 2.0-μM concentrations of peptide were required to elicit significant IFN-γ T cell responses to intermediate target determinants (GAD78–97 and HSPp277), and 7 μM or higher concentrations of peptide were needed to elicit significant responses to the late target determinants (GAD217–236 and insulin9–23), as well as to the control self-Ag MSA560–574 (Fig. 5⇓A). Thus, both the ELISPOT analysis and proliferation assays provide evidence of significant differences in the overall functional avidity of T cells reactive to the early, intermediate, and late β cell target determinants.
Differential avidity of T cells for β cell autoantigen determinants. Ag-primed lymph node T cell responses to different concentrations of Ags were measured by ELISPOT assay. A, Data shown are the mean number of IFN-γ-secreting SFC per 4 × 105 lymph node cells. Mice were tested simultaneously (in triplicate) in at least two separate experiments (n ≥ 8 total for each group). The background level of spontaneous SFC (medium alone) was ≤5. T cell responses to positive controls PPD and anti-CD3 were similar in all groups. B, Data shown are the percentages of high, intermediate, and low avidity T cells in each Ag-specific repertoire (as described in Materials and Methods).
Further analysis of the dose-response ELISPOT data allowed us to determine the relative frequency of high, medium, and low avidity T cells in each Ag-specific T cell pool (Fig. 5⇑B). Notably, the GAD524–543-reactive T cell repertoire, like that of HEL11–25, contained a large proportion of high avidity T cells. Other autoantigen-reactive precursor T cell pools were comprised of very few, or no, high avidity T cells. Instead, the intermediate target determinant-reactive T cell pools consisted of medium-low avidity T cells, and the late target determinant-reactive T cell pools were comprised predominantly of low avidity T cells.
Together, these data show that the tested β cell Ags have had very different impacts on the frequency and the avidity of autoantigen-reactive precursor T cells in preautoimmune NOD mice. Importantly, both the overall functional avidity and the extent to which each autoantigen-reactive T cell pool is comprised of high avidity T cells correlate precisely with the developmental sequence in which spontaneous T cell immunity will arise in these mice.
Discussion
A predefined chronology for the spreading of autoreactivity during the progression of T cell-mediated autoimmune diseases has been well described in animal models and in humans but the basis for this phenomenon is little understood. Notably, relapses of EAE have been associated with the spreading of autoimmunity to specific myelin determinants, and the induction of tolerance to these late target determinants, even after the onset of autoimmunity, can prevent relapsing EAE (6). Accordingly, elucidating the factors that establish the predefined pattern in which T cell autoimmunity spreads may provide insights into disease pathogenesis and aid in the design of stage-specific immunotherapeutics.
T cells specific for foreign Ags often respond to the foreign determinants that have the best MHC binding properties. In contrast, self-reactive T cells are likely to be negatively selected for well presented self-peptides, and those that are allowed to persist may be only weakly responsive to such determinants. This was elegantly demonstrated in studies of T cell repertoires in myelin basic protein (MBP)-expressing and MBP-deficient mice. The peptides MBP78–87 and MBP121–140 are the immune-dominant determinants when MBP is a foreign Ag, in MBP-deficient H-2k and H-2u mice, respectively (26, 27). However, the same determinants cull most of their reactive T cells, and are only weakly immunogenic determinants in H-2k and H-2u MBP-expressing strains. Instead, other determinants of MBP, which have larger pools of reactive T cells available, can more efficiently induce EAE. Little is known about the impact of β cell Ags on the reactive T cell repertoire in animals that are susceptible to spontaneous β cell autoimmunity.
Using a highly sensitive ELISPOT assay and a panel of peptides that included the major target determinants of different β cell autoantigens, we confirmed that autoreactive T cell responses arise in a defined chronological order during NOD mouse development. Next, we showed that the binding affinity of these peptides for I-Ag7 had no relationship with the pattern in which spontaneous autoimmunity spreads. Furthermore, we did not observe a correlation between the affinity of autoantigen peptides for I-Ag7 and the size of their reactive precursor T cell pools in preautoimmune NOD mice. These findings are consistent with previous reports that the ability of autoantigen determinants to prime immune responses in NOD mice is unrelated to their binding affinity for I-Ag7 (28, 29). Thus, unlike immune responses to foreign Ags, the pattern in which spontaneous T cell autoreactivity emerges during the course of the disease is independent of the affinity of the autoantigen determinant for the I-Ag7 molecule.
This lack of correlation between MHC binding affinity and the immunogenicity of β cell Ags (as reflected by the spreading hierarchy) indicates that the preimmune T cell repertoire was shaped by the endogenous self-Ag during tolerance development. To evaluate the impact of each target determinant on the T cell repertoire, we primed preautoimmune mice with an excess of each determinant, thereby activating the full avidity spectrum of reactive T cells, and providing a measure of the clonal size of autoantigen-reactive T cells that escaped negative selection. We found that the clonal size of autoantigen-reactive precursor T cells correlated with the sequence of spontaneous T cell autoimmunity, with the exception of insulin9–23. Although T cell autoreactivity to insulin9–23 appears late, the clonal size of insulin9–23-reactive T cells is significantly larger than that of earlier target determinants.
The differences in the clonal size of precursor T cells responding to autoantigens can partially explain the spreading hierarchy. If the autoreactive T cells were simultaneously primed by endogenous autoantigen, they may become detectable sequentially as the largest, followed by progressively smaller, autoantigen-reactive T cell pools, expanded to detectable levels. However, the precursor frequencies differ by only 2.5-fold between the early and late target determinants, and it is difficult to imagine how this factor alone could account for the 8 wk that separate the appearance of these autoreactivities. Moreover, this scenario does not fully explain the determinant spreading hierarchy, as a complete correlation was only seen after accounting for T cell avidity.
The immunogenicity of self-determinants should also depend on the avidity of reactive precursor T cells. Indeed, the rate of TCR phosphorylation and the subsequent proliferative response of T cells are in part modulated by the avidity of the TCR for the MHC/peptide complex (30, 31). Furthermore, T cell avidity may be an especially critical factor in determining the stage at which T cells reach their activation threshold as the autoimmune process gradually develops in the islets of NOD mice. We found that the precursor T cell pool reactive to the early target GAD524–543, like that of HEL11–25, had high overall avidity, and proportionally was comprised of a large percentage of high avidity T cell clones. Subsequently, T cell autoreactivity spreads to β cell Ag determinants that had progressively lower overall T cell avidity, and whose respective precursor T cell pools were increasingly comprised of low avidity T cell clones. Thus, the frequency of high avidity autoantigen-reactive T cells (in terms of both their actual number and the proportion to which they comprised the Ag-specific repertoire) correlated precisely with the spreading hierarchy of spontaneous T cell autoimmunity. These findings suggest that the extent of selection by self-determinants, which impacts both the size of T cell precursor pools and their avidity, is the major factor responsible for establishing the spreading hierarchy.
Conceivably, the preautoimmune T cell repertoire of NOD mice contains β cell-reactive T cells with a range of different avidities, but which fail to interact with self-Ags at sufficient levels for activation. As an inherent perturbation develops in their islets, the high avidity β cell Ag-reactive T cells, which require less Ag and costimulation, should be the first to exceed their activation thresholds and expand (32, 33). These high avidity T cells should also have faster expansion kinetics (30, 31). This early wave of proinflammatory activity creates a microenvironment (via the local release of cytokines, the recruitment of activated APC pools, and the up-regulation of T cell activation-associated molecules), which supports the activation of lower avidity T cells (1). The activation of the large high avidity T cell response against GAD appears to be crucial for establishing and driving a self-perpetuating autoimmune response, as the early inactivation or circumvention of the GAD-specific T cell response prevents the development of autoimmunity to β cell Ags and insulitis (8, 34). The first wave of autoreactive T cells generates further proinflammatory positive feedback, so that T cells with progressively lower avidities are recruited and participate as effector cells. Such sequential activation of T cells based on their avidity may underlie the observed spreading of β cell autoreactivity in NOD mice during disease progression.
Recently, GAD524–543 was found to contain two determinants, a stronger determinant that is recognized by the spontaneous autoimmune response, and a weaker determinant that is recognized by regulatory T cells (see figures 2⇑ and 4⇑ within Ref. 35). Thus, the large precursor pool reactive to GAD524–543 may reflect reactivity to both of these determinants. However, given that the precursor pool reactive to GAD524–543 is ≈2-fold larger than that to the other tested Ags, the precursor pool reactive to the stronger, spontaneously recognized determinant of GAD524–543 should still be larger than any other tested determinant.
Previous studies of EAE have suggested that the frequency of precursor T cells contributes to the spreading hierarchy (T cell avidity was not evaluated) (6, 36). However, in our spontaneous model of IDDM, the size of the precursor T cell pools did not completely correlate with the spreading hierarchy. Notably, myelin Ags have been shown to extensively shape the T cell repertoire by negative selection, and it appears that low avidity T cells mediate the disease (26, 37, 38). Accordingly, the frequency of myelin-reactive T cells may be the major factor in determining their immunogenicity and encephalogenicity following immunization with myelin Ags. In contrast, NOD mice have deficiencies in tolerance development (12, 39), and many high avidity self-reactive T cells may be allowed to persist. Theoretically, when a perturbation gradually arises in their islets, it is the high avidity T cells that should first become activated and drive subsequent spreading. Thus, in spontaneous autoimmune disease, the avidity of the autoantigen-reactive T cell repertoire may be the most important factor in determining the spreading hierarchy.
In summary, T cell autoreactivity was first detected to a β cell determinant that had both the largest precursor T cell pool and the highest proportion of high avidity T cells. With disease progression, T cell autoreactivity spread to β cell Ag determinants with progressively smaller and lower avidity T cell pools. The hierarchy of determinant spreading partially correlated with the size of precursor T cell pools, and perfectly matched the order of the frequency of high avidity T cells. These features of Ag-reactive T cell repertoires are all shaped by the immunogenicity of β cell Ags during tolerance development. Thus, our findings suggest that the extent to which T cells were negatively selected by the self-determinants is the key factor determining the spreading hierarchy of spontaneous T cell autoimmunity. These results may provide insights into the pathogenesis of organ-specific autoimmune diseases and aid in the design of stage-specific immunotherapies.
Footnotes
-
↵1 This work was supported by grants from the National Institutes of Health, the Juvenile Diabetes Foundation International, and the RIVA Foundation.
-
↵2 Address correspondence and reprint requests to Dr. Daniel Kaufman, Department of Molecular and Medical Pharmacology, University of California, Los Angeles, Medical School, Box 173517, University of California, Los Angeles, CA 90095-1735. E-mail address: dkaufman{at}mednet.ucla.edu
-
↵3 Abbreviations used in this paper: EAE, experimental allergic encephalomyelitis; GAD, glutamic acid decarboxylase; HEL, hen egg lysozyme; HSP, heat shock protein; IDDM, insulin-dependent diabetes mellitus; NOD, nonobese diabetic; SFC, spot-forming colonies; MSA, mouse serum albumin; PPD, purified protein derivative; MBP, myelin basic protein; ELISPOT, enzyme-linked immunospot.
- Received February 5, 2001.
- Accepted April 5, 2001.
- Copyright © 2001 by The American Association of Immunologists
References
In this issue
