Suppressive Effect of Glutamic Acid Decarboxylase 65-Specific Autoimmune B Lymphocytes on Processing of T Cell Determinants Located Within the Antibody Epitope¹

Activated B cells can take up Ag at very low concentrations via their Ag-specific membrane-bound Ig (mIg)³ and process it for presentation to MHC class II-restricted T cells. The enhanced Ag capture by B cells can lower the threshold for a T cell response compared with “professional” APCs (1). B lymphocytes can present Ag to both the Th2 and the Th1 subsets of CD4 T cells (2). The ability of B cells to present minute amounts of Ag may be particularly important for maintaining a state of chronic autoimmunity and inflammation when most of the target tissue has been destroyed. For example, in patients with autoimmune hypothyroidism who have no trace of remaining thyroid gland, immune responses to thyroid peroxidase, an exclusive thyroid Ag, can be detected throughout life (3, 4). In type 1 diabetes, autoimmune responses to the pancreatic β cell autoantigen glutamic acid decarboxylase (GAD) 65, are often maintained for decades following the disappearance of a measurable β cell function (5). Pancreatic β cell destruction resulting in development of type 1 diabetes appears to be mediated by the Th1 subset of CD4 T cells and by CD8 T cells, whereas autoantibodies do not seem to play a role (see Ref. 6 for review). Yet in the nonobese diabetic (NOD) mouse model of type 1 diabetes, clinical signs of β cell failure do not develop in animals carrying a mutation that results in B cell deficiency (7, 8). This protective effect appears to reflect a critical role of B cells as APCs for initiating and maintaining autoimmune T cell responses to pancreatic β cell autoantigens (9, 10, 11). In contrast, skewing of the B cell repertoire toward a β cell autoantigen promotes the development of diabetes in this model (12). While there is evidence to suggest that type 1 diabetes can develop in the face of a severe depletion of mature B cells in man (13), the disease process may be delayed (14).

In addition to the enhanced Ag uptake mediated by mIg of an Ag-presenting B cell, its Ab specificity may influence which T cell epitopes are processed and presented in the context of MHC class II Ags (15, 16). The endocytosed mIg-Ag complex is transported to late endosomal compartments, where the bound Ab can profoundly influence the proteolytic processing and loading onto MHC class II Ags. It has been proposed that an enhancement of the presentation of an epitope can occur when 1) the bound Ig protects it from proteolytic degradation and enhances its survival and therefore availability for binding to MHC class II Ags; 2) the bound Ig facilitates loading of the epitope into the peptide binding groove of MHC-class II Ags; and/or 3) the Ab suppresses presentation of other epitopes, thus increasing the availability of MHC class II molecules for a specific peptide binding in the peptide binding compartment. Conversely, suppression of the presentation of an epitope may result from a prolonged Ag-Fab association in late endosomes preventing processing into the appropriate peptide and/or from a steric blockade of the epitope (16).

The general effect of soluble Abs that form immune complexes with a protein is to enhance Ag uptake via the FcR on macrophages and dendritic cells, resulting in enhanced presentation by such cells. Numerous studies using soluble mAbs have shown that they enhance presentation of the Ag they bind to by facilitating Ag capture through the FcR (see Ref. 17 for review). Furthermore, a study using GAD65 autoantibody-positive sera from type 1 diabetic patients found an enhancement of presentation of a T cell determinant residing in aa 274–286 when GAD65 immune complexes rather than Ag alone were fed to peripheral blood APCs (18). The epitope specificity of a soluble Ab that binds to an Ag before uptake via the FcR can also affect the processing and presentation of T cell epitopes and either suppress or boost a particular epitope (15, 16). While boosting effects may be difficult to separate from the general enhancement of uptake of immune complexes compared with free Ag, relative suppressing effects are more easily demonstrated. For instance, a mouse mAb to thyroglobulin was found to exert a relative suppression on the presentation of a nondominant epitope in this molecule (19). Thus, Ab specificities are likely to modulate the T cell epitopes made available by all three APCs: macrophages, dendritic cells, and B lymphocytes.

Recently, we have mapped in detail the GAD65-specific epitopes recognized by human mAbs derived from type 1 diabetic patients and shown that almost the entire surface of native folded GAD65 is targeted by human autoimmune B cells (20). In this study we assess how the epitope specificities of four corresponding EBV-transformed DRB1*0401-positive B cell lines affect immunodominant T cell epitopes generated from regions within and outside the B cell epitope region. Furthermore, we have studied how soluble Abs from the same B cell lines affect T cell epitopes presented by DRB1*0401-positive splenocytes.

The derivation of EBV-transformed, GAD65-specific, monoclonal B cell lines DPA, DPB, DPC, and DPD from an HLA-DRB1*0401-positive type 1 diabetic patient was described previously (21). The B cell lines have a stable production of IgG1 in tissue culture (21), and the epitopes have been mapped using homologue scanning mutagenesis, taking advantage of the lack of reactivity of the human mAbs with the highly homologous isoform GAD67 (20). The Abs recognize four distinct conformational epitopes in the N-terminal domain (DPB, DPD), middle domain (DPC), and C-terminal domain (DPA) (20) (Fig. 1). Their affinity constant (K_d) for binding to GAD65, determined by surface plasmon resonance on BIAcore, is 0.11 nM for DPA, 9.45 nM for DPB, 0.31 nM for DPC, and 2.79 nM for DPD (A.-M. Madec, unpublished observations). The B cell lines were grown in RPMI 1640 medium (Life Technologies/BRL, Gaithersburg, MD) containing 10% heat-inactivated FCS, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 μM 2-ME (tissue culture medium). Abs were purified from culture supernatant by G4B fast flow (Sigma-Aldrich, St. Louis, MO) affinity chromatography. The eluate was dialyzed against PBS. The human mAb concentration was determined using an IgG ELISA kit (Roche, Mannheim, Germany).

The derivation of HLA-DRB1*0401-restricted, GAD65-specific T cell hybridomas from mice expressing HLA-DRA*0101, DRB1*0401, and human CD4 as transgenes on the I-Ab^−/− background was described previously (22). The hybridomas 91, 103, and 81 recognizing aa 271–285, 481–495, and 556–570 in GAD65, respectively, were used in this study. A fourth HLA-DRB1*0401-restricted, GAD65-specific T cell hybridoma, recognizing aa 115–130 (23), was donated by Dr. L. Wicker (Sharp and Dohme Research Laboratories, Merck, Rahway, NJ). T cell hybridomas were maintained in tissue culture medium (see above).

Human GAD65, expressed in Saccharomyces cerevisiae and purified as described (24), was donated by Drs. M. Powell, B. Rees-Smith, and J. Furmaniak (FIRS Laboratories, Cardiff, U.K.). The purified protein was conformationally stable and was recognized by a series of human mAbs recognizing conformational epitopes in GAD65, including the mAbs used in this study (results not shown).

EBV-transformed, GAD65-specific B cell lines were incubated with graded amounts of Ag for 1 h on ice and then washed twice in tissue culture medium to remove unbound Ag. Ag-loaded B cells (3 × 10⁵) were added to each well of a round-bottom 96-well plate containing 1.5 × 10⁵ GAD65-specific T cell hybridomas in a final volume of 150 μl tissue culture medium. The cells were incubated for 24 h (37°C, 4% CO₂). The HLA-DRB1*0401-positive human EBV-transformed B cell line, Priess (25), which does not express a GAD65-specific IgG was used as a negative control. IL-2 secretion was measured using a sandwich immunoassay and a streptavidin-europium detection system (26). Experiments were repeated a minimum of three times in triplicate wells.

Preformed immune complexes were generated by incubating 1.5 μg/ml purified human mAbs and graded amounts of Ag (0.03–3.0 μg/ml) in a round-bottom 96-well plate for 1 h at room temperature before addition of APCs. Splenocytes were prepared from HLA-DR A1*0101^+/+, B1*0401^+/+, hCD4^+/+, I-Ab⁻/⁻ transgenic mice (23) and used as APCs. Splenocytes (3 × 10⁵ cells/well in 50 μl tissue culture medium) were incubated with graded amounts of Ag alone, peptide alone, or the preformed immune complexes for 1 h before addition of T cell hybridomas (1.5 × 10⁵ cells in 100 μl tissue culture medium were added per well). After 24-h incubation, the samples were processed for measurement of IL-2 secretion as described above.

The rat anti-mouse FcR mAb 2.4G2 (IgG2b) was used to block FcRII receptor uptake of immune complexes. The mAb was purified from supernatants of HB197 cells (American Type Culture Collection, Manassas, VA) using G4B Fast Flow (Sigma) column affinity chromatography and added to splenocyte experiments (10 μg/ml) using preformed immune complexes as Ag.

The four EBV-transformed monoclonal HLA-DRB1*0401-positive monoclonal B cell lines used in this study recognize four distinct conformational epitopes residing on the surface of the GAD65 molecule (20). Fig. 1 shows the linear and spatial relationships between the B cell epitopes of these four lines compared with four immunodominant GAD65 T cell epitopes recognized by HLA-DRB1*0401-restricted T cell hybridomas (22, 23). DPB and DPD B cells recognize distinct conformational epitopes in the N-terminal domain of GAD65, which were mapped to aa 39–173 and 96–173, respectively (20). The determinant of the T35 T cell hybridoma (aa 115–130) overlaps or is adjacent to the DPB and DPD epitopes. The third B cell line, DPA, recognizes a conformational epitope residing in the C-terminal domain of GAD65, involving residues in aa 483–499 and 556–585. In particular, H568 is critical for binding (20). The epitopes of two of the T cell hybridomas, HY103 (aa 481–495) and HY81 (aa 556–570), overlap or are proximal to the DPA epitope. The fourth B cell line, DPC, recognizes a conformational epitope residing in the middle domain of GAD65 and involving aa 231–234 and 366–413. The epitope of the fourth T cell hybridoma, HY91 (aa 271–285), localizes to the same domain, but does not overlap with the DPC epitope (Fig. 1). Thus, in contrast to the DPA, DPB, and DPD epitopes, the DPC epitope does not overlap with a T cell determinant recognized by the panel of T cell hybridomas. In the three dimensional model of the middle and C-terminal regions of GAD65 (20), the HY103 and HY81 epitopes are exposed on the surface of the GAD65 dimer, while the HY91 epitope is at the interface between the middle and C-terminal domains, which form a sandwich in each monomer (Fig. 2). It is therefore buried in the native GAD65 molecule. A three-dimensional model of the N-terminal region is not available. However, in the predicted secondary structure of GAD65 (20) the determinant of the T35 T cell hybridoma (aa 115–130) resides in an amphipathic α helix, the hydrophilic phase of which is likely to be exposed on the surface of native GAD65 (20).

We evaluated the ability of the DPA, DPB, DPC, and DPD B cell lines to process and present GAD65 to the four T cell hybridomas used in this study (Fig. 3). The Priess B cell line, which shares the HLA-DRB1*0401 haplotype with the DP-B cells, but does not recognize GAD65 epitopes, was used as a control APC. All the DP-B cell lines elicited similar IL-2 responses from the DRB1*0401-restricted, GAD65-specific T cells used in this study when synthetic GAD65 peptides were used as an Ag, indicative of a similar expression level of HLA-DRB1*0401 on their surface. The Priess cell line presented synthetic GAD65 peptides to DRB1*0401-restricted GAD65-specific T cells more efficiently than any of the DP-B cell lines (results not shown), consistent with a higher expression level of HLA-DRB1*0401 on its surface.

The Priess cell line has been shown to nonspecifically (pinocytosis) take up GAD65 at high concentrations and process it for presentation to T cells (27) (Fig. 3). Because of the enhanced uptake of GAD65 via mIg, the DP cell lines would be expected to process and present whole GAD65 more efficiently than Priess unless the bound Ig exerts an inhibiting effect on a particular epitope. Consistent with this prediction, the DPC cell line, which binds to a conformational epitope distant from all the T cell epitopes recognized by our panel of T cell hybridomas, elicited severalfold higher IL-2 levels than Priess from the four T cell hybridomas following incubation with whole GAD65 (Fig. 3 and Table I). Similarly, DPA, DPB, and DPD elicited severalfold higher IL-2 responses than Priess from T cell hybridomas recognizing T cell epitopes outside their Ab binding region (Fig. 3 and Table I). By contrast, the presentation of T cell epitopes residing in or adjacent to the Ab epitope was either blocked or inhibited. Thus, presentation of the aa 556–570 peptide, which is within the Ab epitope of DPA, was efficiently blocked when DPA was used for Ag presentation (Fig. 3,B). Similarly, presentation of another T cell epitope (aa 481–495), which resides within or adjacent to the DPA Ab epitope, was inhibited compared with presentation of this same peptide by the other B cell lines and was only slightly enhanced compared with that of Priess (Fig. 3,C). In contrast, DPA efficiently presented the N-terminal and middle domain epitopes (aa 115–130 and 271–285, respectively) to the relevant T cells (Fig. 3, A and D, and Table I).

Similar results were obtained with the DPB and DPD B cell lines. DPB presented the T cell epitope 115–130, which resides within or adjacent to its Ab epitope, at a similar level as Priess, and presentation was effectively inhibited compared with that of DPA and DPC (Fig. 3,D). Yet DPB was able to present the middle and C-terminal domain epitopes efficiently (Fig. 3, A–C, and Table I). Similarly, presentation of the 115–130 epitope by DPD was moderately improved compared with Priess, but poor compared withDPA and DPC (Fig. 3,D). In contrast, DPD presented the middle and C-terminal domains at similar levels as DPC and severalfold more efficiently than Priess (Fig. 3, A–C, and Table I).

In summary, the autoimmune GAD65-specific B cell lines appear to efficiently present T cell epitopes residing outside their Ab epitope region, while presentation of T cell determinants that reside within the Ab-bound region of GAD65 is suppressed in a dominant fashion (Table I).

We next analyzed whether the inhibiting effect of the DPA, DPB, and DPD B cell lines on processing and presentation of T cell epitopes located within their Ab-bound GAD65 regions was also relevant when immune complexes were taken up, processed, and presented by APCs like macrophages and dendritic cells. For these experiments purified GAD65 or GAD65 in an immune complex with DPA, DPB, DPC, or DPD Abs was incubated with HLA-DRB1*0401-positive splenocytes, which were then used to stimulate the T cell hybridomas. The binding of immune complexes by the FcRII receptor on macrophages enhances their uptake compared with Ag alone (17).

Compared with incubation with Ag alone, the incubation of splenocytes with GAD65 in an immune complex with DPA, DPB, DPC, or DPD Abs resulted in a severalfold enhanced stimulation of the HY91 hybridoma, recognizing the 271–285 peptide (Fig. 4,A). Similarly, the presentation of peptides 115–130, 481–495, and 556–570 was severalfold enhanced following incubation of splenocytes with GAD65 in an immune complex with purified DP Abs, which do not recognize epitopes that overlap with those peptides (Fig. 4). In contrast, loading of a complex between GAD65 and DPA Ab, which recognizes an epitope overlapping with the 481–495, and 556–570 peptides, resulted in an inhibition of their presentation to levels similar to that with Ag alone (Fig. 4, B and C), with a more complete inhibition of epitope 556–570. Likewise, loading of a complex between GAD65 and either DPB or DPD Abs, which recognize B cell epitopes overlapping with or adjacent to the 115–130 epitope, resulted in an inhibition of its presentation compared with DPA and DPC Abs, with DPB Ab inhibiting presentation more completely than DPD (Fig. 4 D). Thus, purified DPA, DPB, and DPD Abs in an immune complex with GAD65 inhibited presentation of T cell epitopes by macrophages in a similar manner as when the corresponding B cells were used for Ag presentation.

Abs enhance uptake via the FcRII on macrophages. To confirm that the stimulating effect on outlying epitopes observed in the experiments with soluble Abs is dependent on an enhanced uptake via the FcRII molecule, the experiments described above were repeated in the presence of an FcRII Ab, which blocks uptake by the FcIIR, but not by pinocytosis. In the presence of the FcRII Ab, none of the DP Abs in an immune complex with GAD65 enhanced presentation to T cell hybridomas compared with Ag alone (Fig. 5 and results not shown), suggesting that the general boosting effect of soluble DPA, DPB, DPC, and DPD on T cell epitopes outside the Ab binding site is the result of increased Ag capture via the FcRII.

The results presented here suggest that autoimmune GAD65-specific B cells and the Abs they secrete may play a major role in shaping the autoimmune T cell response by specifically suppressing presentation of T cell determinants residing within Ab-bound regions. Ab effects on uptake are generally positive whether expressed on the surface of Ag-presenting B cells or forming soluble immune complexes that are taken up by FcR on APCs. However, the expected increase in Ag presentation is only observed for T cell determinants located outside the region captured by Ab. In contrast, T cell determinants located within or adjacent to the B cell epitope are specifically suppressed, and the suppressive effect dominates. One possible explanation for this effect is that binding of a GAD65 autoantibody to its epitope is stable throughout the acidic environment of Ag processing and MHC class II loading compartments, resulting in masking of T cell determinants residing in the region. While the binding affinities of the DP Abs varied 86-fold between DPA (highest) and DPB (lowest; see Materials and Methods), both mAbs and the corresponding cell lines effectively blocked the presentation of T cell determinants located within the Ab epitope. Thus, T cell epitope suppression appears to occur over a wide range of Ab affinities.

The findings of epitope suppression are consistent with earlier studies in which tetanus toxin-specific mIg as well as the corresponding soluble Abs suppressed the processing and presentation of T cell determinants within the footprint of the Ab (15, 16). In previous studies a detailed mapping of the B cell epitopes in tetanus toxin was not available. Instead, the suppressed T cell determinants were found to be localized within the Ab footprint, i.e., the Ag fragment that is protected by Ab during limited proteolysis and usually encompasses a larger area than the Ab epitope. In one case, the Ab footprint included both a T cell epitope that was suppressed and an epitope that was enhanced by Ab binding (16). Based on these results, Simitsek et al. (16) proposed a model in which the processing and MHC class II loading of a T cell epitope residing in the actual Ab contact site (epitope) are suppressed, while the loading of a nearby epitope residing inside the footprint but outside the actual binding site is favored. The present study has identified three cases of effective suppression of immunodominant DRB1*0401-restricted T cell determinants residing within or adjacent to regions constituting the epitope of human monoclonal GAD65 autoantibodies arising in type 1 diabetes. We did not identify a T cell epitope whose processing/presentation were relatively enhanced by Ab binding to the epitope area, but the possibility is not excluded by this limited set of B cell lines and T cell hybridomas.

Both immunization of DRB1*0401-positive transgenic mice (22, 23) and derivation of human T cell lines from human patients (28) have identified the aa 271–285 determinant as perhaps the most immunodominant DRB1*0401-restricted T cell epitope in the GAD65 molecule. In the three-dimensional model of the middle and C-terminal regions of GAD65, the 271–285 epitope is buried in the native folded molecule. This is in contrast to both the aa 481–495 and 556–570 T cell epitopes, which are exposed on the surface. More than 95% of type 1 diabetic patients exclusively make autoantibodies to conformational epitopes on the surface of the native GAD65 molecule (Ref. 20 and references therein). The aa 271–285 T cell epitope does not, therefore, appear to reside in the Ab binding site of a typical autoimmune B cell in type 1 diabetes. For example, while the epitopes of two other human mAbs derived from diabetic patients, MICA 6 and MICA 10, have been mapped close to the 271–285 T cell epitope (242–282 region), both Abs bind the surface of native intact GAD65, suggesting that the buried 271–285 residues are not part of the epitope. Furthermore, in experiments using splenocytes as APCs, MICA 10 in an immune complex with GAD65 does not suppress presentation of the 271–285 determinant (W. Richter, J. C. Jaume, S. L. Parry, G. Sønderstrup, and S. Baekkeskov, unpublished observations). We hypothesize that the absence of B cell reactivity to the aa 271–285 region, and the consequent absence of suppression of processing and presentation of T cell determinants in this region by B cells and APCs may contribute to the immunodominance of this epitope.

Taken together, our results suggest that autoimmune GAD65-specific B cells not only serve a critical function as APCs in type 1 diabetes, but also play a major role in shaping the epitope specificity of the autoimmune T cell response to GAD65. Thus, autoimmune GAD65-specific B cells and the Abs they secrete appear to modulate the autoimmune T cell repertoire by down-regulating T cell epitopes in an immunodominant area while boosting epitopes in distant regions, providing a mechanism for autoimmune T cell epitope spreading.

We are grateful to Drs. M. Powell, B. Rees-Smith, and J. Furmaniak (FIRS Laboratories) for donation of purified GAD65 and to Dr. L. Wicker (Merck, Sharp and Dohme) for donation of T-35.