Abstract
Natural development of diabetes in nonobese diabetic (NOD) mice requires both CD4 and CD8 T cells. Transgenic NOD mice carrying αβ TCR genes from a class I MHC (Kd)-restricted, pancreatic β cell Ag-specific T cell clone develop diabetes significantly faster than nontransgenic NOD mice. In these TCR transgenic mice, a large fraction of T cells express both transgene derived and endogenous TCR β chains. Only T cells expressing two TCR showed reactivity to the islet Ag. Development of diabetogenic T cells is inhibited in mice with no endogenous TCR expression due to the SCID mutation. These results demonstrate that the expression of two TCRs is necessary for the autoreactive diabetogenic T cells to escape thymic negative selection in the NOD mouse. Further analysis with MHC congenic NOD mice revealed that diabetes development in the class I MHC-restricted islet Ag-specific TCR transgenic mice is still dependent on the presence of the homozygosity of the NOD MHC class II I-Ag7.
Insulin dependent diabetes mellitus is caused by an autoimmune response to the pancreatic β cell Ag(s) (1). During the course of the disease, both cellular and humoral responses to an array of β cell Ags have been demonstrated (1). However, the precise mechanisms by which β cells are destroyed are not well understood. The nonobese diabetic (NOD)3 mouse develops diabetes spontaneously and shares many of the key features of the human disease (2). The development of disease in the NOD is characterized by an initial cellular infiltrate into the pancreatic islets (insulitis) which causes the gradual loss of pancreatic β cells resulting in the development of hyperglycemia. Disease can be transferred from diabetic NOD mice to young nondiabetic NOD mice by lymphocytes but not by Abs, and transfer requires both CD4 and CD8 T cells (3). Furthermore, depletion of either CD4 or CD8 T cells in vivo protects NOD mice from development of the disease (4, 5, 6). However, under certain experimental conditions, either CD4 and CD8 T cell clones specific for the islet Ag(s) can transfer diabetes (7, 8). Furthermore, transgenic mice carrying either CD4 or CD8 islet Ag-specific TCR develop diabetes in the absence of endogenous TCR expression (SCID and RAG−/− backgrounds) (9, 10). These results suggest that islet Ag-specific CD4 or CD8 T cells alone can cause diabetes. Thus, although the requirement for both CD4 and CD8 T cells is well established for the natural development of diabetes, the particular stage at which CD4 and CD8 T cells cooperate is still unclear, and the precise mechanisms by which T cells destroy pancreatic β cells remain largely unknown.
During development, T cells undergo positive and negative selection in the thymus (11). These selection events ensure the generation of mature peripheral T cells capable of responding to foreign Ag-derived peptides bound to the self MHC complex and elimination of self-reactive T cells. However, in the NOD mouse, both class I and class II MHC-restricted, self Ag-reactive T cells escape negative selection in the thymus, emigrate to the periphery, and are activated to differentiate into diabetogenic effector T cells. For CD4 T cells, we have demonstrated a link between the unique biochemical nature of NOD class II MHC, I-Ag7, and poor negative selection in the thymus (12, 13). However, it is still not known how self-Ag-specific (islet Ag), class I MHC-restricted diabetogenic CD8 T cells are generated in the thymus. Furthermore, the underlying events leading to the activation of self-Ag-specific CD8 T cells are not well understood. In this report, using transgenic NOD mice carrying islet Ag-specific, class I MHC-restricted TCR gene, we demonstrate that the expression of two TCRs during thymic selection is required for the generation of self-reactive T cells in the thymus and that this mechanism plays an important role in the generation of islet-Ag-specific diabetogenic CD8 T cells in the periphery. We also show that the diabetes development in the islet Ag-specific class I-restricted TCR transgenic mice is influenced by class II MHC gene.
Materials and Methods
Mice
NOD and NOD.SCID mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and maintained at the Washington University School of Medicine animal facility under specific pathogen-free conditions. NOD.GD mice carrying Kd, I-A αd/βd, no I-E and Db, MHC genotype was established in our animal facility by introducing MHC locus form B10GD to NOD mice (14 generation backcross).
Establishment of NOD transgenic mice carrying islet Ag-specific class I MHC-restricted TCR
The 9.33 T cell clone recognized murine β cells expressing Kd, class I MHC molecules and exhibit cytolytic activity (14). The 9.33 clone was positively stained with an anti Vβ6 mAb and express in-frame Vβ6-Dβ1-Jβ1.4-Cβ1 mRNA. Two α-chain mRNAs were found in the clone, a Vα19 and a second from a previously undescribed Vα family, referred to as Vα clone 6. The Vα19 gene was rearranged to the Jα32 segment. The junctional sequence of this cDNA was in-frame, and the mRNA encoded a functional TCR α-chain. Sequence analysis of the Vα clone 6 revealed an out-of-frame joining. Genomic fragments of each receptor chain were cloned. A clone containing Vα/Jα was further subcloned into a TCR α shuttle vector (provided by Dr. M. M. Davis, Stanford University, Stanford, CA) (15) using XhoI/NotI restriction sites. The VDJ fragment of TCR β-chain was subcloned into a TCR β-chain shuttle vector (also provided by Dr. M. M. Davis) (15) using ClaI/NotI restriction sites. A mixture of TCR α- and β-chain constructs was injected into fertilized eggs from NOD mice, and the injected egg was implanted into pseudo-pregnant hosts. Mice were screened by PCR analysis of tail biopsy DNA using TCR α- and β-specific primers (Vα, GCCATGAAAACATATGCTCCTACA and Jα, CAGAGAACTGAGCCTAATG; and Vβ, CCAAACTATGAACAA and Jβ, CGCCCAGTTCCCCGT) (data not shown). Positive offspring were tested for the expression of transgene derived TCR using anti-Vβ6 mAbs. Two independent founders were identified and gave identical results. The founder mice were mated with NOD mice, and offspring were screened for the presence of the transgene using anti-Vβ6 mAb staining of peripheral blood lymphocytes.
9.33 TCR transgenic mice were mated with NOD.GD (H-2g6) mice, and the resulting F1 mice were backcrossed to NOD.GD mice to produce 9.33 TCR-positive NOD mice with H-2g7/g7, H-2g7/g6, and H-2.g6/g6 mice. In the MHC heterozygous mice, expression of I-Ag7 and I-Ad class II MHC molecules are equivalent as demonstrated in the previous study (16).
Diabetes
Mice were monitored for the development of diabetes by measuring urine glucose with Chemistrip (Boehringer Mannheim, Indianapolis, IN) twice a week. Mice with glucosuria were tested for the blood glucose level, and those showing >250 mg/dl of blood glucose for two consecutive readings in a week were considered diabetic.
Abs and surface immunofluorescence staining
mAbs used in this study are as follows: anti-CD4 (GK1.5 and RL172.4), anti-CD8 (3.155), anti-TCR β-chain (H57), and a series of anti-TCR V β-chain Abs (KJ25:Vβ3, MR9-4:Vβ5, RR4-7:Vβ6, MR10-2:Vβ9, MR11-1:Vβ12, and MR12-4:Vβ13) were produced in our laboratory and biotynilated. PE-conjugated anti-CD4 Ab (H129.19) and FITC-conjugated anti-CD8 Ab (53-7.3) were purchased from PharMingen (San Diego, CA). Cells were stained with a combination of Abs described in the text with appropriate secondary reagents as described previously (17). Stained cells were analyzed by FACScan using the CellQuest software (Becton Dickinson, San Jose, CA). Sorting of stained cells were conducted using FACSVantage fluorescence-activated cell sorter as described (17).
Proliferation assay and the establishment of islet Ag-specific cell line
Pancreatic islet cells were prepared by the methods described previously (14). Responder cells (5 × 104) were incubated with an indicated numbers of irradiated (2000 rads) islet cells in a final volume of 200 μl 5% FCS DMEM in flat-bottom 96-well microtiter plates. Cultures were harvested after 5 days of incubation with a 6-h pulse with 1 μCi of [3H]thymidine. Islet-reactive CD8-positive T cells were established by culturing CD4-depleted lymph node cells from 9.33 TCR transgenic mice (1 × 105) with 2 × 103 irradiated islet cells in 200 μl 5% DMEM in U-bottom 96-well microtiter plates. After 1 wk of incubation, recovered cells (2 × 104) were stimulated with irradiated NOD spleen cells (2.5 × 105) and 2 × 103 irradiated islet cells in the same fashion and maintained by weekly stimulation with Ag. Spleen cells (2 × 105/ml) from NOD mice or Vβ6-positive T cells obtained by FACS sorting (5 × 104/ml) and stimulated with Con A (1 μg/ml) in the presence of rIL-2 (25 U/ml) were used as control.
Diabetes transfer
Transfer of diabetes by spleen cells was conducted by the method described previously (18). In brief, spleen cells (2 × 107) from diabetic mice were injected i.v. into irradiated (650 rads) young prediabetic NOD male mice. Mice were monitored for the development of diabetes for 3 wk. For the depletion of T cell subsets, spleen cells were treated two cycles by either anti-CD4 (RL172.4) or anti-CD8 (3.155) Ab plus complement. This treatment depletes >99% of specific T cell subset.
Results
Accelerated development of diabetes and presence of islet Ag-specific CD8 T cells in 9.33 TCR transgenic NOD mice
TCR VαJα- and VβDβJβ-chain genes were cloned from the 9.33 islet Ag-specific CD8 T cell clone (14) and inserted into TCR transgenic vectors containing the constant region of the TCR and other regulatory elements known to direct specific expression of transgenic TCR in vivo. TCR α and β constructs were coinjected into NOD fertilized eggs. Offspring were initially identified for the presence of transgene. Transgene-positive mice were further screened for the expression of transgenic TCR β-chain (Vβ6) on peripheral blood T lymphocytes. Two founder mice were identified and used for the subsequent experiments. Mice were mated to NOD mice, and transgene-positive offspring were screened by the same method. All female transgenic offspring from both founders developed diabetes significantly earlier than nontransgenic (Fig. 1⇓). The control mice developed diabetes with similar incidence and kinetics to the original NOD mice obtained from The Jackson Laboratory (Fig. 1⇓). The rapid development of diabetes made it difficult to use female transgenic mice for breeding, and therefore only male 9.33 TCR transgenic NOD mice were used for breeding. However, male transgenic mice also developed diabetes significantly faster than normal NOD male with higher penetrance (100% of transgenic mice were diabetic at 35 wk compared with only 25% diabetes incidence in nontransgenic littermate, data not shown).
Incidence off diabetes in female 9.33 TCR transgenic and control nontransgenic littermate NOD mice. TCR transgenic and nontransgenic NOD mice (15 mice for each group) were monitored for the development of diabetes as described in Materials and Methods.
CD4 and CD8 T cells from lymph nodes of 3 wk old (pre-diabetic) 9.33 TCR transgenic mice were tested for their islet Ag reactivity in vitro using pancreatic β cells prepared from NOD.SCID mice (Table I⇓). CD8-positive lymph node cells but not CD4-positive cells from 9.33 TCR transgenic mice proliferated to the islet Ag in a dose-dependent manner. The CD8-positive islet Ag-reactive T cell lines established from the 9.33 TCR transgenic mice exhibited cytolytic activity to pancreatic islet cells expressing Kd class I MHC Ag as demonstrated with the original 9.33 T cell clone (14). Neither CD8- nor CD4-positive lymph node T cells from age-matched nontransgenic NOD mice responded to islet Ag in vitro. These results are in agreement with the report by Verdaguer et al. (10) and demonstrate the presence of transgene TCR-bearing islet Ag-specific CD8 T cells in 9.33 TCR transgenic NOD mice that accelerate the development of diabetes in the NOD mice.
Islet Ag-specific response of 9.33 TCR transgenic T cellsa
Expression of transgene-derived TCR β-chain in 9.33 TCR transgenic mice
Thymocytes and lymph node T cells from 9.33 TCR transgenic mice were analyzed for the T cell development and the expression of transgene derived TCR β-chain. Unlike other transgenic mice carrying CD8 T cell-derived TCR genes (19, 20), no skewing to CD8 single positive (SP) T cell subset was evident among the thymocytes. The lymph node T cells also showed no skewing to CD4 or CD8 SP T cell subset distribution (Fig. 2⇓A). A majority of thymocytes expressed Vβ6 TCR chains at a high or low level and some were negative (Fig. 2⇓B). These populations represent the developmental stages of the thymocyte, from CD4 CD8 double negative (DN) TCR-negative T cell precursor, immature CD4 and CD8 double positive (DP) TCRlow thymocytes, and mature CD4 or CD8 SP TCRhigh T cells.
Lymphocytes subset distribution and transgenic TCR β-chain expression in 9.33 TCR transgenic mice. A, Thymocytes and lymph node cells from 9.33 TCR transgenic mice were stained with anti CD4 PE and anti-CD8 FITC and analyzed for the subset distribution as described in Materials and Methods. B, Thymocytes from 9.33 TCR transgenic mice were stained with anti-Vβ6 Ab (bold line) or none (solid line) followed by FITC-labeled goat anti-rat Ig Ab. Stained samples were by FACScan using CellQuest software.
The analysis of transgenic TCR expression on peripheral mature T cells showed marked heterogeneity among T cells from 9.33 TCR transgenic mice. As shown in Fig. 3⇓, all of the T cells from transgenic mice were uniformly stained with anti-TCR β-chain Ab, but the expression of Vβ6 varied among them with high and intermediate and negative cells. This heterogeneity was primarily represented in the CD8 T cell subset. CD4 T cells showed uniform staining with anti-Vβ6 Ab (Fig. 4⇓). Analysis of the transgenic TCR α-chain expression was conducted by RT-PCR using PCR primer corresponding to the Jα-Vα sequence of the transgene. Abundant message was found in both thymus and peripheral T cell populations. However, surface expression of the transgenic TCR α-chain could not be monitored due to the lack of TCR α-chain-specific Ab. This TCR expression pattern was present in both transgenic founder mice, ruling out the founder effect for this phenotype.
Expression of transgene-derived Vβ6 TCR in peripheral lymph node T cells. Lymph node cells from normal NOD mice and 9.33 TCR transgenic mice were incubated with anti-Thy-1 Ab and counterstained with none (solid line), biotynilated anti-pancreatic-TCR β-chain Ab (dotted line), or biotynilated anti-Vβ6 Ab (bold line), washed, and then incubated with FITC-goat anti-rat Ig Ab and avidin-PE as described in Materials and Methods. Thy-1-positive populations were analyzed for the expression of TCR β-chain and transgene-derived Vβ6 chain.
Expression of transgene-derived Vβ6 TCR on CD4 and CD8 subpopulation of T cells in 9.33 TCR transgenic mice. Lymph node cells from 9.33 TCR transgenic mice were stained with either anti-CD4 or anti-CD8 Abs and counterstained with none (solid line), anti-pancreatic TCR β-chain Ab (dotted line) of anti-Vβ6 Ab by the methods described in Fig. 3⇑.
Presence of T cells expressing two TCR β-chains and their reactivity to the islet Ag
The presence of intermediate Vβ6 expressing CD8-positive T cells in 9.33 TCR transgenic mice prompted us to examine whether these T cells were expressing two different TCR β-chains on their cell surface. T cells were stained with anti-Vβ6 Ab and counterstained with a mixture of anti-VβAbs (anti-Vβ3, -5, -8, -9, -12, and -13). Staining of the cells from normal NOD mice showed an absence of T cells expressing two TCR β-chains (Fig. 5⇓), confirming a strict allelic exclusion for the expression of TCR β-chain. In contrast, a large number of T cells from 9.33 TCR transgenic mice expressed both transgene-derived Vβ6 and other TCR β-chains (Fig. 5⇓). It should be noted that analysis of TCR Vβ-chain usage among Vβ6 intermediate T cell populationx showed no dominant endogenous Vβ-chain usage, indicating a polyclonal nature of this double TCR β-chain expressing T cells (data not shown). Thus, it seems that allelic exclusion for TCR β-chain expression is not operational in this TCR transgenic mouse line.
Expression of two TCR β-chains on T cells from 9.33 TCR transgenic mice. Lymph node cells from normal NOD mice and 9.33 TCR transgenic mice were incubated with anti-Vβ6 Ab and a mixture of biotinylated anti-Vβ Abs washed and then incubated with FITC goat anti-rat Ig and avidin-PE as described in Materials and Methods. Stained samples were analyzed by FACScan using CellQuest software.
To test the function of the CD8-positive T cells with double TCR β-chain expression, cells were sorted based on their surface Vβ6 TCR expression and stimulated with pancreatic β cells. As shown in Table II⇓, Vβ6 intermediate T cells mounted a significant proliferative response to pancreatic islet cells in an Ag dose-dependent fashion. No proliferative response was found in the T cells expressing high level of Vβ6 TCR on the surface.
Islet Ag-reactivity of TCR intermediate T cells from 9.33 TCR transgenic micea
In a different experiment, CD4-depleted lymph node cells from 9.33 TCR transgenic mice were stimulated with pancreatic β cells, and the bulk CD8 T cells lines were then examined. As shown in Fig. 6⇓, Con A-activated normal NOD lymph node cells contained both Vβ6-positive and other TCR expressing T cells but did not have T cells coexpressing Vβ6 and other TCRs on their surface. Moreover, T cells from NOD mice selected for the expression of Vβ6 were all positive with anti-Vβ6 Ab and did not express any other TCR Vβ-chains. In contrast, a large proportion of the islet reactive T cells from 9.33 TCR transgenic mice stained both with anti-Vβ6 Ab and with a mixture of anti-TCR Vβ Abs. This result suggests that the expression of two Vβ-chains on their cell surface (Fig. 6⇓). It should be noted that the mixture of Abs used in this study stain ∼40% of all freshly isolated T cells and about 30% of islet reactive Vβ6-positive T cells. Thus the distribution of non-Vβ6 TCR chain expression on the islet Ag-specific Vβ6-positive T cell population was similar to that of normal T cell population in nontransgenic NOD mice. These results clearly demonstrate that islet reactive T cells in 9.33 TCR transgenic mice were contained in the T cell population expressing both transgenic TCR Vβ-chain and endogenous Vβ-chains.
Expression of two TCR β-chains on islet reactive T cells derived from 9.33 TCR transgenic mice. Con A-stimulated lymph node T cells from NOD mice, islet Ag-specific T cell line from 9.33 TCR transgenic mice, and anti-Vβ6 stimulated lymph node T cells from normal NOD mice were stained by the same method described in Fig. 5⇑.
Lack of T cell development in 9.33 TCR transgenic mice in NOD.SCID background
To test whether the expression of two TCR β-chains is necessary for the generation of mature 9.33 TCR-positive T cells and for the development of diabetes, the 9.33 TCR transgenic mice were crossed onto the NOD.SCID mice. Mice were screened for the lack of B cells and for the presence of transgene as described in the previous section. Female mice were monitored for the development of diabetes (Table III⇓). At the end of the experiments (35 wk of age), neither transgene positive nor control NOD.SCID mice developed diabetes.
Lack of diabetes development in 9.33 TCR transgenic NOD.SCID micea
Thymocytes and peripheral lymph node cells from TCR transgenic NOD.SCID mice were analyzed for the development of T cells (Fig. 7⇓). The number of thymocyte in 9.33 TCR transgenic NOD.SCID mice (4- to 6-wk-old mice) ranged from 23 to 30 × 106. This is significantly more than that of nontransgenic NOD.SCID mice (10–15 × 106 cells/thymus), but far less than that of age-matched normal 9.33 TCR transgenic mice (120–160 × 106 cells/thymus). Staining of 9.33 TCR transgenic NOD.SCID thymocytes with anti-CD4 and anti-CD8 Abs showed decrease of DP and SP cells. Furthermore, emigration of mature T cells from thymus to the peripheral lymphoid organs was drastically reduced, and only few T cells were present in the lymph node of the 9.33 TCR transgenic NOD.SCID mice (Fig. 7⇓). These results suggest that the absence of endogenous TCR due to the SCID background inhibited the maturation of 9.33 TCR-bearing T cells and consequently impeded the development of diabetes.
Analysis of thymocyte and lymph node cells from 9.33 TCR transgenic SCID mice. Thymocytes from 9.33 TCR transgenic NOD.SCID and wild-type NOD mice were stained with anti-CD4 and anti-CD8 Abs as described in Fig. 1⇑. Lymph node cells were stained with FITC-labeled anti-Thy-1 Ab followed by goat anti-rat Ig Ab.
Role of MHC class II gene in the development of diabetes in 9.33 TCR transgenic mice
Lack of T cell development in 9.33 TCR transgenic SCID mice did not allow us to examine whether 9.33 TCR-positive CD8 T cell alone can cause diabetes in vivo. Therefore, we used the T cell transfer system to examine the diabetogenecity of CD8 T cells from 9.33 TCR transgenic mice (Table IV⇓). CD4- or CD8-depleted spleen cells from diabetic 9.33 TCR transgenic NOD mice were transferred to sublethally irradiated prediabetic NOD male mice. Mice that received neither CD4- nor CD8-depleted spleen cells developed diabetes, but those that received a mixture of spleen cells developed diabetes within 3 wk after cell transfer (Table IV⇓). This result suggests that, although accelerated, diabetes development still needs both CD4 and CD8 T cell subsets in 9.33 TCR transgenic NOD mice.
Diabetes transfer by 9.33 NOD T cellsa
In separate experiments, we examined the role of the class II MHC, I-Ag7, in the development of diabetes in 9.33 TCR transgenic mice. The TCR transgene was introduced onto a recently established NOD.GD mouse line. This MHC congenic NOD mouse line expresses Kd, I-Ad, and Dd MHC molecules and differs from NOD mouse only at class II I-A gene expression (I-Aαdβd instead of diabetes susceptible NOD class II MHC I-A αdβg7). None of NOD.GD and (NOD.GD × NOD)F1 female mice developed diabetes within 35 wk of observation. Two of five (NOD.GD × NOD)F1 mice carrying 9.33 TCR transgene developed diabetes with significantly delayed kinetics compared with NOD 9.33 TCR transgenic mice. The five NOD.GD mice carrying the 9.33 TCR transgene also remained free of diabetes (Table V⇓).
Diabetes development in 9.33 TCR transgenic NOD MHC congenic micea
The TCR expression on CD4 and CD8 peripheral T cells from NOD.GD 9.33 TCR transgenic mice were similar to the 9.33 TCR NOD mice in that all of CD4 T cells expressed transgene-derived Vβ6 TCR, while the expression of Vβ6 TCR on CD8 T cell was heterogeneous, with cells expressing high, low, and no Vβ6 on the surface (data not shown). Islet Ag reactivity of the CD8 T cells from NOD.GD 9.33 TCR transgenic mice was tested in vitro. The CD8 T cells from TCR transgenic NOD.GD mice, but not from the wild-type NOD.GD mice, mounted strong proliferative response to islet cells prepared from NOD.GD mice (Fig. 8⇓). The CD8 T cells from 9.33 TCR transgenic NOD.GD mice were similar to those observed in the 9.33 NOD mice (Table I⇑ and Fig. 8⇓). Thus, it seems that NOD.GD 9.33 TCR transgenic mice possess islet Ag-reactive transgenic TCR-positive CD8 T similar to that of NOD 9.33 TCR transgenic mice. However, class II genotype, either I-Ad homozygosity, I-Ad/I-Ag7 heterozygosity, or I-Ag7 homozygosity, significantly influenced the development of diabetes in this TCR transgenic mouse line.
Islet Ag reactivity of T cells from 9.33 NOD.GD mouse. CD4-depleted lymph node cells (5 × 104 cells/well) from 9.33 NOD.GD (hatched bar) and from wild-type NOD.GD mice were stimulated with NOD.GD islets (3 × 103 cells/well) by the method described in Table I⇑.
Discussion
The transgenic mice established in this study, in agreement with the similar study conducted by Verdaguer et al. (10), clearly demonstrate that introduction of TCR transgene from islet Ag-specific CD8 T cell clone significantly accelerates the disease progression in NOD mice. This is likely due to the increased number of islet Ag-specific CD8 T cells in these transgenic mouse lines (Table I⇑). The onset of diabetes in our TCR transgenic mouse line was similar to the previous report (10). However the incidence of diabetes in our transgenic mouse line was 100% in both male and female, whereas 80% of female and 40% of male transgenic mice established by Verdaguer developed diabetes. This difference is likely due to the effect of the background genes on the development of diabetes in two different TCR transgenic mouse lines. We have introduced the TCR genes directly into NOD mouse fertilized egg to establish transgenic mouse line with homogeneous genetic background. In contrast, Verdaguer et al. (10) used (B6 × SJL)F2 eggs for TCR gene injection, and the resulting transgenic mice were backcrossed onto the NOD mice. Although the known idd genetic loci that influence the development of diabetes can be monitored and screened out during backcross, these backcrossed mice may still differ at unidentified genetic loci that influence diabetes development as shown previously (21). It should be noted that during backcrossing, mice developing diabetes before reproduction are naturally eliminated, and hence simple backcrossing inevitably introduce the selection of a viable phenotype. Thus, this possible genetic heterogeneity among the TCR transgenic mice generated by backcrossing may account for the difference in the development of diabetes in the previous study (10). The transgenic mouse line established in our study is free of these potential problems, and this genetic homogeneity accounts for the reproducible development of diabetes among our TCR transgenic mice.
Analysis of transgenic TCR expression on the total T cell population as well as islet Ag-specific CD8-positive T cells from 9.33 TCR transgenic mice revealed that a majority of islet Ag-specific CD8 T cells express both transgenic TCR and endogenous Vβ-chains. In contrast, islet Ag nonreactive CD8 T cells and all of the CD4 T cells seem to express a single TCR Vβ-chain. T cell development in 9.33 TCR transgenic mice was severely impaired on the NOD.SCID background at the DP stage, and no mature T cells emigrated to the periphery. These results strongly indicate that T cells expressing transgenic TCR are capable of interacting with Ag(s) expressed in the thymus, and this interaction impedes T cell development in the absence of endogenous TCR rearrangement in SCID mice. However, in the presence of endogenous TCR β-chain rearrangement, T cells expressing two TCR β-chain (transgene-derived Vβ and endogenous Vβ-chains) are capable of escaping this negative influence in the thymus. At present we do not know whether the ligand expressed in the thymus is the same as the ligand expressed in pancreatic β cell to which the 9.33 T cell clone was originally raised. However, it has been demonstrated that certain pancreas-specific Ags, such as insulin, glucagon, and pancreatic polypeptide, are also expressed in the thymus, albeit at a significantly lower level than in the pancreas (22). It is possible that the 9.33 T cell clone may be specific for one of these Ags expressed in both the thymus and the pancreas and thus can be negatively selected on the SCID background. Identification of the antigenic peptide and the pancreatic Ag recognized by the 9.33 T cell clone would be necessary to examine the possible expression of this diabetogenic Ag in the thymus.
The possibility that self-reactive T cells escape by expression of two TCRs was suggested by a study of T cells from TCR transgenic mice (23). Similar to our 9.33 TCR transgenic mice, T cells expressing self C5-specific transgenic TCR were detected only in mice capable of rearranging endogenous TCR chains but not in the RAG−/− mice. This finding was further extended by Sarukhan et al. (24) using a combination of three transgenic mouse lines expressing hemagglutinin (HA) specific, class II MHC-restricted TCR transgene, transgenic HA expression driven by rat insulin promoter (pancreatic islet-specific expression) and expression of transgenic HA in the thymus. It was shown that T cells expressing HA-specific TCR can escape thymic negative selection by expressing two TCR α-chains and are capable of causing diabetes in the mouse expressing HA in pancreatic β cells. These results clearly established that the lack of allelic exclusion in TCR α-chain gene (expression of two TCRs) plays an important role in the escape of autoreactive and pathogenic T cells from the thymic-negative selection. However, these results were obtained under experimental conditions generated in mice carrying multiple transgenes. Our results, using transgenic mice carrying TCR from islet Ag-specific T cell clone, are clearly the first demonstration that in spontaneously developing autoimmune diabetes, pathogenic T cells escape from thymus-negative selection by expressing two TCRs and these T cells are capable of accelerating the disease process.
It has been shown that allelic exclusion mechanism tightly regulates TCR β-chain expression and normal T cells express a single TCR β-chain (25). In contrast, a majority of T cells exhibit rearrangement of TCR α-chain gene on both alleles and a significant number of T cells express two TCR α-chains on their surface associated with a single TCR β-chain (26, 27). Thus it was proposed that lower expression of self-reactive TCR due to competition from the second TCR α-chain for the TCR β-chain plays an important role for the escape of potentially autoreactive T cells from thymic-negative selection (24, 28). However, in 9.33 TCR transgenic mice, the expression of endogenous TCR β-chain seems to play a major role in decreasing the expression of islet Ag-specific TCR. We do not know whether this lack of allelic exclusion of TCR β-chain is a general phenomenon for the diabetogenic CD8 T cells in NOD mouse or a unique feature of the 9.33 TCR transgenic mice. It has been shown that NOD mouse carrying transgenic TCR β-chain from either OVA-specific I-Ad-restricted CD4 T cells (29) or islet Ag-specific Kd-restricted CD8 T cell clone (30) showed accelerated development of diabetes. In view of our finding that expression of two TCR β-chain may be required for the escape of diabetogenic CD8 T cells, these transgenic TCR β-chains may rescue T cells expressing diabetogenic TCR from thymic-negative selection via similar mechanisms observed in our 9.33 TCR transgenic mice. Thus it is possible that expression of two TCR β-chains may play a critical role in the generation of a pathogenic CD8 T cell repertoire and the development of diabetes in NOD mice.
There was no evidence for negative selection of diabetogenic CD8 T cells in the thymus of TCR transgenic mice reported previously (10). The possible difference in TCR affinity to the Ags between two islet Ag-specific CD8 T cell clones may account for the disparity in their development in the transgenic mouse thymus. Alternatively, these CD8-positive islet Ag-specific T cells recognize different Ags and the expression of these Ags in the thymus may differ resulting in different developmental pattern in different TCR transgenic mice. Thus, it would be of interest to examine an extended number of islet-specific CD8 T cell clones for their Ag specificity, capacity to accelerate diabetes development in NOD mouse, and susceptibility to the thymic negative selection in vivo.
The second unique feature of our TCR transgenic mice is a CD4 T cell dependency for the development of diabetes in 9.33 TCR transgenic mice. The failure of disease transfer by CD8 T cells from diabetic 9.33 TCR transgenic mice clearly demonstrate that in this transgenic mice as well as normal NOD mouse, both CD4 and CD8 T cells are required for the diabetes development in the recipients. Our study with NOD.GD mice also demonstrated that acceleration of diabetes development in 9.33 TCR transgenic mice depends on the class II MHC genotype of the NOD mice. The lack of diabetes development cannot be accounted for by the absence of islet Ag-specific CD8 T cells, because CD8 T cells from 9.33 TCR transgenic NOD.GD mice vigorously respond to islet Ag stimulation in vitro. These findings show that cooperation of diabetogenic CD4 T cells restricted to I-Ag7 and islet Ag-specific 9.33 TCR-positive CD8 T cells are required for the diabetes development.
Our findings are somewhat different from that observed in the previous islet Ag-specific class I-restricted TCR transgenic mice (10). In that study, there was little effect of MHC heterozygosity on the diabetes development. In 9.33 TCR transgenic mice, due to the unique selection of 9.33 TCR-positive diabetogenic T cells in the thymus, only a fraction of CD8-positive T cells are capable of responding to islet Ag (Fig. 5⇑). These islet Ag-reactive T cells express two TCRs resulting in lower Ag-specific TCR expression and lower interaction affinity to the target pancreatic β cells. Under this condition, similar to the normal development of diabetes in NOD mice, generation of effector CD8 T cells may require help from I-Ag7-restricted islet Ag-specific CD4 T cells. The presence of I-Ad class II MHC disrupts this necessary cellular cooperation and protects mice from diabetes in both 9.33 TCR transgenic NOD.GD and normal NOD.GD mice. Alternatively, initial interaction between CD8-positive T cells and islet cells may need to recruit CD4-positive islet Ag-specific T cells to initiate the diabetogenic process as demonstrated by class I MHC-deficient NOD mice (6). This process, especially the generation of islet Ag-specific I-Ag7-restricted CD4 T cells, may be influenced by class II MHC heterozygosity and would explain the reduced incidence of diabetes in 9.33 TCR transgenic (NOD × NOD.GD)F1 mice. Nevertheless, a combination of the islet Ag-specific class I-restricted TCR transgenic mice and class II MHC congenic NOD mouse line, NOD.GD, would provide a useful tool to examine the cooperativity of two types of T cells as well as the requirement for MHC class II I-Ag7 homozygosity in the development of diabetes in the NOD mouse model of insulin dependent diabetes mellitus.
Acknowledgments
We thank E. R. Unanue, A. Suri for critically reading the manuscript, M. White for microinjection, K. Frederick for help with maintaining mouse colonies, and M. M. Davis for TCR transgene expression constructs.
Footnotes
-
↵1 This work was supported by grants from the Juvenile Diabetes Foundation, the National Institutes of Health, and the Kilo Diabetes and Vascular Research Foundation.
-
↵2 Address correspondence and reprint requests to Dr. Osami Kanagawa, Department of Pathology, Washington University Medical School, Box 8118, 660 South Euclid Avenue, St. Louis, MO 63110. E-mail address: kanagawa{at}pathbox.wustl.edu
-
↵3 Abbreviation used in this study: NOD, nonobese diabetic; DN, double negative; DP, double positive; SP, single positive; RAG, recombination-activating gene.
- Received November 16, 1999.
- Accepted March 1, 2000.
- Copyright © 2000 by The American Association of Immunologists
References
In this issue
