Cutting Edge: Thymic Positive Selection and Peripheral Activation of Islet Antigen-Specific T Cells: Separation of Two Diabetogenic Steps by an I-A^g7 Class II MHC β-Chain Mutant¹

Insulin-dependent diabetes mellitus (IDDM)³ is a T cell-mediated autoimmune disease under the complex regulation of both genetic and environmental factors (1, 2). In both human IDDM and its mouse model, the NOD mouse, specific allelic forms of the class II MHC genes are the most important genetic elements predisposing to diabetes development (3, 4) The IDDM-susceptible class II MHC genes, DQ2 and DQ8 in the human and I-A^g7 in the NOD mouse, share a unique nonaspartic acid residue at position 57 of their β-chains (5, 6) Transgenic expression of a modified I-A^g7 (I-A^g7PD), in which histidine and serine residues at position 56 and 57 of I-A^g7 β-chain were replaced by proline and aspartic acid, respectively, protected NOD mice from developing diabetes (7, 8). How these specific residues of class II MHC molecules modulate the development of diabetes is not well understood. Here, we show that I-A^g7PD class II MHC molecules are capable of positively selecting diabetogenic CD4 T cells in the thymus, but fail to present pancreatic β cell Ag to activate the same T cells in the periphery.

cDNA encoding I-A^g7 β-chain with proline and aspartic acid at positions 56 and 57 (9) was introduced into a transgenic expression vector under the control of an I-E class II MHC-specific promoter (10). The transgenic DNA construct was injected into fertilized BALB/c eggs, and mice were screened for surface expression of transgene on peripheral blood lymphocytes using the I-A^g7-specific mAb, 10.3.6.2. Transgenic mice carrying islet Ag-specific TCR, BDC2.5, were described previously (11). BDC2.5-transgenic mice were mated to CB17.SCID mice to produce BDC2.5 H-2^d SCID. BALB/c and NOD mice were purchased from The Jackson Laboratory (Bar Harbor, ME). (NOD × BALB/c)F₁ mice were bred in our colony. All mice were kept under specific pathogen-free conditions in the Washington University School of Medicine animal facility.

Spleen cells (1 × 10⁶/sample) were incubated with biotinylated anti-I-A^d (MKD6), anti-I-A^g7 (10.3.6.2) mAbs, or without Ab at 4°C for 25 min, then washed and reincubated with streptavidin-coupled FITC (Caltag, South San Francisco, CA) for an additional 25 min. Samples were analyzed with FACScan using CellQuest software (Becton Dickinson, Mountain View, CA).

T cells (2 × 10⁴) were cultured with 2.5 × 10⁵ irradiated (2000 rad) spleen cells in the presence of the indicated dose of Ag in a final volume of 200 μl 5% FCS in DMEM in flat-bottom microtiter plate. For stimulation of spleen cells with islet cells, spleen cells (1 × 10⁵) from bone marrow chimeric mice were stimulated with irradiated NOD spleen cells, as described above. Cultures were harvested after a 72-h incubation with a 6-h pulse with [³H]thymidine.

Bone marrow cells from BDC2.5 H-2^d SCID mice were treated with anti-Thy1 mAb (At83 1/10 v/v) and complement for 45 min at 37°C. T cell-depleted bone marrow cells (1 × 10⁷) were injected into irradiated (850 rad) recipients. Chimeric mice were analyzed 8 to 10 wk after reconstitution.

BALB/c transgenic mice (named BALB^g7PD) expressing the introduced I-A^g7PD β-chain with the same tissue distribution as the endogenous I-A^d class II molecules (data not shown) were selected for the experiments. Spleen cells from BALB^g7PD, NOD, BALB/c, and (NOD × BALB/c)F₁ mice were stained with mAbs specific for I-A^g7 and I-A^d β-chains (10.3.6.2 and MKD6, respectively). As shown in Figure 1, the cells from (NOD × BALB/c)F₁ mice express equivalent amounts of I-A^d and I-A^g7. The I-A^g7 and I-A^d molecules differ only in their β-chains and share an I-A α^d-chain (5). The equivalent expression of the two β-chains on the cell surface of (NOD × BALB/c)F₁ mice demonstrates that I-A^d and I-A^g7 β-chains have a similar affinity for the I-A^d α-chain. However, in BALB^g7PD mice, the expression of the transgene-encoded Aβ appears slightly lower than that of endogenous I-A^d class II MHC molecules (Fig. 1). We found the same staining pattern in two independent founders. We know that all of the mAb specific for the I-A^g7 β-chain interact with residues surrounding positions 56 and 57 (12). Moreover, in transfected B cell lines expressing only I-A^g7PD (I-Aα^d and β^g7PD), the cell surface staining for the complex appears lower with the anti β-chain Ab than with the anti α-chain Ab (E. Carrasco-Marin, unpublished observation). Therefore, it is likely that the lower staining intensity of I-A^g7PD on transgenic spleen cells is due to the lower affinity of the mAb to the modified I-A^g7 β-chain.

To assess the function of mutant I-A^g7PD on the APCs, we measured their Ag-presenting capacity using three OVA-specific T cell clones (Fig. 2). The OVA-2 clone recognizes OVA peptide (323–339) presented by either I-A^d or I-A^g7, as demonstrated in our previous report (13). This cell line, as expected, responded to OVA Ag presented by splenic APC from NOD, BALB/c, (NOD × BALB/c)F₁, and BALB^g7PD mice. The OVA-3 clone responds to the same peptide presented only by I-A^g7 (13) and, in contrast, failed to respond to OVA Ag presented by BALB^g7PD splenic APC. A third clone, OVAPD, was established from BALB^g7PD mice and showed no reactivity to the OVA Ag presented by I-A^d class II MHC. However, this clone did respond to OVA presented by NOD, (NOD × BALB/c)F₁, and BALB^g7PD spleen APC (Fig. 2). Thus, the unique I-A^g7PD β-chain associated with endogenous I-Aα^d-chain was expressed on the transgenic mouse APC and was capable of presenting protein Ag to T cells.

The presentation of islet β cell Ag to T cells by the I-A^g7PD class II molecule was tested using the diabetogenic CD4 T cell clone BDC2.5 (11, 14, 15). The BDC2.5 T cell clone, established from diabetic NOD mice, responds to pancreatic β cell Ag presented by the I-A^g7 class II MHC molecules. This diabetogenic T cell line responded to BALB/c-derived islet cells in the presence of (NOD × BALB/c)F₁ APC (Fig. 3). In this assay, Ag from the islets are transferred to the APC expressing class II MHC for presentation to the BDC 2.5 T cell. Moreover, the presentation of islet Ag to BDC2.5 T cell is I-A^g7 restricted, because non-NOD APC such as BALB/c fail to present islet Ag (14). Importantly, no proliferative response was observed to BALB/c islet cells in the presence of APC from the BALB^g7PD transgenic mouse (Fig. 3). These results indicate that the modification of I-A^g7 at positions 56 and 57 of the β-chain completely abolishes its capacity to present islet Ag to the diabetogenic BDC2.5 T cells.

The BDC2.5 TCR transgene was introduced into the CB17.SCID strain (H-2^d) (16). The thymocytes from BDC H-2^d SCID mice were analyzed by surface immunofluorescence (Fig. 4, Donor). A majority of thymocytes were arrested at the immature CD4/CD8 double-positive stage and failed to produce mature CD4 single-positive T cells. These results establish that I-A^d is not a positively selecting MHC for BDC2.5 TCR-bearing thymocytes, while as previously established, the I-A^g7 is the selecting I-A allele (15).

Bone marrow cells from BDC H-2^d SCID mice (Donor in Fig. 4) were transferred into lethally irradiated (850 rad) (NOD × BALB/c)F₁, BALB/c, and BALB^g7PD transgenic mice. Eight weeks after bone marrow reconstitution, thymocytes from the bone marrow chimeras were analyzed (Fig. 4). As expected, a large number of CD4 single-positive T cells (32% of total thymocytes) were generated in the (NOD × BALB/c)F₁ thymus, whereas no mature single-positive T cells were found in the thymus of BALB/c recipients. In the BALB^g7PD recipient thymus, the mature CD4 single-positive T cells (26% of total thymocytes) were generated as efficiently as in the (NOD × BALB/c)F₁ recipient thymus, indicating that BDC2.5 TCR-bearing T cells were positively selected by I-A^g7PD class II MHC. It should be noted that a majority of the thymocytes in all three different recipients expressed Vβ4 TCR (the TCR V β-chain used by the BDC2.5 T cell) (11), indicating that these thymocytes were derived from BDC TCR transgenic bone marrow cells (data not shown). Furthermore, having used the BDC H-2^d SCID mice as bone marrow donors, we excluded the development of T cells using endogenous TCR (16).

None of the chimeric mice developed diabetes over a 3-mo period. This was not surprising given that all of the hemopoietic cells, including all of the APC, in the chimeric mice are derived from BDC2.5 H-2^d SCID bone marrow cells and are, by virtue of their H-2^d MHC expression, incapable of presenting diabetogenic Ag to BDC.2.5 T cells. However, the peripheral T cells in the chimera mice were functional because they responded to islet Ag in an I-A^g7 MHC-restricted fashion as demonstrated in an in vitro proliferation assay (Fig. 5). We recently found that the mice carrying both BDC2.5 TCR and I-A^g7PD transgenes on the CB17 SCID background also showed no development of diabetes (result not shown). In these mice, BDC2.5 TCR-positive CD4-positive T cells developed efficiently, similar to the born marrow chimera shown in this report.

In summary, the results presented here demonstrate that thymocytes bearing the BDC2.5 TCR interact with the mutant I-A^g7 molecule (I-A^g7PD) in the thymus and that this interaction transduces positive selecting signals. In contrast, no productive interaction takes place between the mature BDC2.5 TCR-positive T cells and the mutant I-A^g7 molecule and islet Ag(s). These results help us to interpret previous studies in which an I-A^g7 β-chain with the PD changes was introduced as a transgene into NOD mice (7, 8). These mice did not develop diabetes. We would conclude that the protection was not caused by a negative effect of mutant I-A^g7 on the selection of diabetogenic CD4 T cells in the thymus. It should be noted that the I-A^g7 molecule as well as I-A^g7PD is a class II MHC molecule that binds peptide poorly (9). This results in a poor negative selection process (17) and could also promote peripheralization of a wide range of T cells. We previously showed that the I-A^g7PD mutation affected either recognition by T cells or peptide binding when testing different T cell clones to different peptides (13). We could not determine which circumstance applied in the case of islet Ags, since we did not know their chemical identity. Thus, the lack of islet Ag-specific activation of BDC2.5 T cell by I-A^g7PD APC could be explained by 1) a lack of peptide binding or 2) poor or low affinity interaction of the BDC2.5 TCR with I-A^g7PD-peptide complex.

A similar dissociation between thymic selection and peripheral activation of T cells has been reported in the class I MHC system by Ohashi et al. (18). T cells bearing lymphocytic choriomeningitis virus (LCMV)-specific, class I D^b molecule-restricted TCR were positively selected in a D^bm13 class I mutant mouse. However, mature T cells in the D^bm13 mouse failed to respond to LCMV antigenic peptide. Thus, both wild-type D^b and mutant D^bm13 were sufficient for positive selection of the transgenic TCR bearing T cells in the thymus, but only the wild-type MHC bound antigenic peptides for activation of the mature T cells. Our findings and those of Ohashi et al. (18) indicate the great plasticity of TCR/MHC/peptide interaction, as well as a lack of absolute correlation between positively selecting MHC/peptide and activating MHC/peptide complex for a single TCR. The biologic significance of these findings in the establishment of T cell Ag repertoire awaits further investigation.

We thank Michael White for production of the transgenic mice. We thank Dr. Charles Kilo and the Kilo Diabetes and Vascular Research Foundation for their generous support.