Autoreactive Diabetogenic T-Cells in NOD Mice Can Efficiently Expand From a Greatly Reduced Precursor Pool

NOD/Lt mice were maintained in a specific pathogen-free research colony at The Jackson Laboratory (Bar Harbor, ME). Currently, type 1 diabetes develops in 90% of female and 63% of male NOD/Lt mice by age 1 year. NOD.AI4αβ transgenic (Tg) mice that transgenically express the TCR from a K^d class I−restricted β-cell autoreactive CD8 T-cell clone have been previously described (21). Mice with a mixed C57BL/6 and DBA/2 genetic background, including an H2^b MHC haplotype, and expressing a transgenic TCR-αβ chain complex specific for a D^b class I−restricted LCMV peptide (officially designated as B6,D2-TgN[TcrLCMV]327Sdz, and here designated as B6,D2.LCMV TCR Tg mice) have also been previously described (23) and were obtained from The Jackson Laboratory’s induced mutant resource. An N9 backcross stock of NOD.LCMV TCR Tg congenic mice fixed to homozygosity for linkage markers delineating all known Idd loci of NOD origin was produced by our previously described methods (26). During the course of establishing this congenic stock, backcross segregants carrying the TCR-α chain transgene were identified by screening genomic DNA prepared from peripheral blood leukocytes (PBL) by polymerase chain reaction (PCR) with the primer pair 5′-CTGACCTGCAGTTATGAGGACAGCAC-3′ and 5′-CGAGGATCCTTTAACTGGTACACAGCAGG-3′, which amplifies a 302-bp product. Presence of the TCR-β chain transgene was determined by PCR with the primer pair 5′-CATGGAGGCTGCAGTCACCC-3′ and 5′-GTTTGTTTGCGAGCTCTGTTTTGATGGCTC-3′, which amplifies a 410-bp product. The previously described stock of T-cell− and B-cell−deficient NOD-scid mice (official designation NOD-Prkdc^scid) is maintained at the N11 backcross generation (27). These latter two strains served as progenitors for a stock designated NOD-scid.LCMV TCR Tg. NOD-scid homozygotes were outcrossed to NOD.LCMV TCR Tg mice. The subsequent F1 progeny were backcrossed to NOD-scid to produce mice homozygous for scid and heterozygous for LCMV TCR Tg. Homozygous scid segregants were identified by the absence of B-cells among PBL using the previously described flow cytometric techniques (26). Some studies used a previously described stock (16) of NOD-scid mice whose pancreatic β-cells express a rat insulin promoter−regulated transgene that encodes the B7-1 T-cell co-stimulatory molecule (NOD-scid.RIPB7). All mice were allowed free access to food (Agway Diet NIH 31A; Agway, South Henly, MO) and acidified drinking water. NOD-scid stocks received s trimethoprim-sulfamethoxazole supplement (Sulfatrim; Barre-National, Baltimore, MD) in the drinking water on an alternative weekly basis.

Diabetes development in the indicated mice was defined by glycosuric values ≥3, as assessed with Ames Diastix (kindly supplied by Miles Diagnostics, Elkhart, IN).

Splenic leukocytes from the indicated 8-week-old female mice were assessed for CD4 and CD8 T-cell levels, and the proportion of these that expressed the transgenic LCMV TCR was determined by multicolor flow cytometric techniques using the Cell Quest 3.3 (FACScan; Becton Dickinson, San Jose, CA) data reduction system. Total T-cells were detected with the TCR-αβ−specific monoclonal antibody H57-597 conjugated to a green fluorescent fluorescein isothiocyanate (FITC) tag. Total T-cells were then further characterized for CD4 expression using the monoclonal antibody GK1.5 conjugated to the red fluorescent tag Cy3.18-OSu (Cy3; Biological Detection Systems, Pittsburgh, PA), or for CD8 expression with the monoclonal antibody 53-6.72 conjugated to phycoerythrin (PE), whose red fluorescence intensity can easily be distinguished from that of Cy3. A separate aliquot of splenic leukocytes from each mouse was assessed by two-color fluorescence-activated cell sorter (FACS) analysis for the total proportion of T-cells expressing the α and/or β chain of the LCMV TCR by respective use of PE- and FITC-conjugated antibodies specific for TCR Vα2 (B20.1) and Vβ8 (F23.1) elements. Specific expression of the transgenic LCMV TCR on CD4 or CD8 T-cells was determined by two-color FACS analysis using FITC-labeled CD4 or CD8 specific antibodies in conjunction with the PE-labeled TCR Vα2 specific antibody. Proportions of CD4 or CD8 T-cells expressing a transgenic or nontransgenic TCR were determined by three-color FACS analysis using APC-labeled CD4 (RM4-5) or PE-labeled CD8 (53-6.72)-specific antibodies in conjunction with FITC-conjugated antibodies specific for TCR Vβ2 (B20.6), Vβ4 (KT4), Vβ5.1,2 (MR9-4), Vβ6 (RR4-7), Vβ7 (TR310), Vβ8.1,2,3 (F23.1), Vβ9 (MR10-2), Vβ10 (B21.5), Vβ11 (RR3-15), Vβ12 (MR11-1), Vβ13 (MR12-3), or Vβ14 (14-2[5]) elements. Our previously described multicolor flow cytometric technique (28) was used to determine proportions of thymic, single positive CD4 and CD8 T-cells expressing TCR-αβ complexes, as well as those expressing the Vα2 component of the LCMV TCR Tg.

Splenic CD4 T-cells were purified from the indicated donors using a streptavidin-conjugated magnetic bead system (Miltenyi Biotec, Auburn, CA) to deplete CD8 T-cells, macrophages/granulocytes, and B-cells that had been prestained with a cocktail of biotinylated antibodies directed against lineage-specific markers. CD8 T-cells and the macrophage/granulocyte populations were depleted with the monoclonal antibodies 53-6.72 and M1/70, respectively , whereas B-cells were removed with a goat polyclonal antiserum specific for mouse Ig molecules (Sigma, St. Louis, MO). Splenic CD8 T-cells were purified by the same negative selection approach, but using a biotinylated antibody specific for CD4 (GK1.5) rather than CD8 in the depletion cocktail. FACS analyses indicated the purity of these CD4 and CD8 T-cell preparations was consistently >95%. Aliquots of 5 × 10⁶ donor CD4 and CD8 T-cells were then injected intravenously alone or in combination into the indicated types of 4- to 6-week-old female NOD-scid recipients. Recipients were then monitored for type 1 diabetes development for 17 weeks.

β-cell autoreactive CD8 T-cells were propagated from islets of the indicated donors by our previously described techniques (16). Briefly, T-cell infiltrated donor islets were layered on top of preseeded lawns of irradiated (2000 R) NOD-scid.RIPB7 islets in tissue culture medium containing interleukin (IL)-2, IL-7, and the monoclonal antibody GK1.5 to block the outgrowth of CD4 T-cells. T-cell outgrowths from these cultures were further expanded by a 1-week passage on NOD-scid.RIPB7 islets, and subsequently phenotyped by the flow cytometric analysis methods described above. In addition, it was also determined what proportion of the cultured CD8 T-cells (identified with the FITC-labeled monoclonal antibody 53-6.72) expressed a TCR capable of recognizing the K^d-bound NRP-A7 peptide (KYNKANAFL), which is a mimotope of a β-cell autoantigen frequently targeted by MHC class I−restricted diabetogenic T-cells in NOD mice (20). This was done by co-staining with a PE-conjugated K^d/NRP-A7 tetramer complex synthesized by previously described methods (29). The soluble K^d MHC class I construct used to generate these tetramers has also been previously described (30).

AI12.B1.3 is a previously described H2-K^d MHC class I−restricted β-cell autoreactive T-cell clone of NOD origin (16). The AI12.B1.3 clone is characterized by a TCR Vα17 to Jα42 gene rearrangement event that is a prevalent feature of diabetogenic CD8 T-cells in NOD mice (15,16). To create a permanent cell line expressing the AI12.B1.3 TCRs, full-length cDNAs encoding its α and β chains were amplified by PCR and cloned into expression vectors as previously described (31). Briefly, the TCR-α chain was cloned into the XhoI/XbaI sites of the expression vector pSH-xs, in which the hygromycin resistance gene originally present in the vector was replaced by a puromycin resistance gene from pPgk-Puro (32). The TCR-β chain was cloned into the XhoI/BamHI sites of the expression vector pcDL-SRα296 (33). The TCR⁻ T-cell hybridoma variant 58α⁻β⁻ (34), engineered by Chang et al. (35) to ensure expression of CD8α, CD8β, and CD3ζ, was generously provided by H.-C. Chang (Dana Farber, Boston, MA). 58α⁻β⁻ cells (1 × 10⁷) were electroporated with linearized AI12.B1.3 TCR-α and -β chain expression constructs (5 and 30 μg, respectively, giving an α:β molar ratio of 1:9 using a Bio-Rad Gene Pulser (0.33 kV, 960 μF). At 48 h after electroporation, cells were seeded in 24-well plates at 3 × 10⁴/well and selected with 2 μg/ml puromycin. Growing clones were analyzed for TCR expression by FACS analysis using an antibody to CD3ε (145-2C11; Pharmingen). Recognition of synthetic peptides by AI12.B1.3 TCR transfectants was measured by IL-2 production, as previously described (31), except that RMA-S/K^d cells (kindly provided by M. Bevan, University of Washington, Seattle, WA) incubated at 31°C overnight were used to present exogenously added synthetic peptides.

A wide range of β-cell autoreactive CD4 T-cells, and a somewhat less diverse, but still broad set of CD8 T-cells, normally contribute to the development of type 1 diabetes in NOD mice. However, the question remained whether a β-cell autoreactive T-cell repertoire sufficient to elicit overt type 1 diabetes would develop under conditions that would greatly limit their precursor frequency. To initially answer this question, we asked whether the allelic exclusion induced by an H2-D^b class I−restricted transgenic TCR, specific for a β-cell−irrelevant LCMV antigen, blocked type 1 diabetes development in NOD mice. This was clearly not the case because, as shown in Fig. 1, at the N9 backcross generation, the rate and frequency of type 1 diabetes development in female NOD.LCMV TCR Tg mice was identical to that of transgene-negative segregants. Furthermore, the rate and frequency of type 1 diabetes in the LCMV TCR Tg−positive and −negative backcross segregants was identical to that of standard NOD/Lt mice housed at The Jackson Laboratory. This latter result provided a functional confirmation that all Idd susceptibility loci necessary for type 1 diabetes development had been fixed in the N9 congenic stock of NOD.LCMV TCR Tg mice.

Our previous studies using another transgenic TCR indicated allelic exclusion might not be an efficient process in NOD mice (21). Thus, it was possible that the high rate of type 1 diabetes development in NOD.LCMV TCR Tg mice was attributable to the fact that these mice continued to develop a significant number of nontransgenic T-cells. To address this issue, we compared the proportion and types of splenic T-cells in which the LCMV TCR Tg was and was not expressed in 8-week-old NOD mice versus in the B6,D2 stock in which it was originally introduced.

As expected based on previously published results (23), the vast majority of total splenic T-cells (∼92%) in B6,D2.LCMV TCR Tg mice expressed the transgenic TCR (Table 1). Also as expected, most T-cells (∼92%) in B6,D2.LCMV TCR Tg mice were in the CD8 compartment. Of these B6,D2 CD8 T-cells, ∼97% expressed the Vα2 component used by the LCMV TCR Tg. Among the few residual CD4 T-cells that remained present in this B6,D2 stock, ∼24% expressed the Vα2 component characterizing the LCMV TCR Tg. In contrast, the proportion of total T-cells expressing the LCMV TCR Tg in NOD mice (∼80%) was lower than that observed in the B6,D2 stock. This primarily resulted from the fact that the NOD mice generated a much larger total number of mostly transgenic negative (∼75%) CD4 T-cells than did B6,D2 mice (Table 1). Such numerical increases in both the LCMV TCR Tg−positive and −negative subsets were attributable to background gene effects, given that significantly higher numbers of CD4 T-cells were also found in standard NOD than in B6,D2 control mice. Nonetheless, the total number of nontransgenic CD4 T-cells in the NOD.LCMV TCR Tg stock (7 × 10⁶) was much less than the total number of CD4 T-cells in standard NOD mice (28 × 10⁶). Hence the pool of CD4 T-cells from which diabetogenic effectors could presumably be drawn was approximately four times less in NOD.LCMV TCR Tg than in standard NOD mice.

As a result of the differential expansion in the CD4 compartment, the overall proportion of T-cells skewed toward the CD8 compartment by the LCMV TCR Tg in NOD mice (∼82%) was also less than that observed in the B6,D2 stock (Table 1). The proportion of NOD CD8 T-cells expressing the LCMV TCR Tg (∼94%) was only slightly less than that of similar cells in B6,D2 control mice. As a result, the total number of nontransgenic CD8 T-cells in the NOD.LCMV TCR Tg stock (3 × 10⁶) was much less than the total number of CD8 T-cells in standard NOD mice (14 × 10⁶). Thus, similar to the case in the CD4 compartment, the pool of CD8 T-cells in which diabetogenic effectors could presumably reside was four to five times smaller in NOD.LCMV TCR Tg than in standard NOD mice.

Another potentially important outcome of these analyses is that, in both the CD4 and CD8 compartments, much higher total numbers of LCMV TCR Tg−positive as well as −negative T-cells accumulated in NOD than in B6,D2 mice (Table 1). The fact that the T-cells that preferentially accumulated in NOD mice include a TCR transgenic clonotype, which presumably would not have encountered their cognate LCMV peptide, indicated that this phenotype is driven in an antigen-independent fashion. Also supporting this conclusion was that analyses of thymuses from 8-week-old NOD.LCMV TCR Tg female mice (n = 3) revealed that the proportions of transgenic CD4 and CD8 single positive T-cells (29.4 ± 2.6 and 90.8 ± 2.0%, respectively) were quite similar to that observed in spleen. Hence the NOD genetic background supported equivalent peripheral expansion of both LCMV TCR Tg−positive and −negative T-cells. Collectively, these data indicate that, although the frequency of any given diabetogenic T-cell clonotype is likely to be less in the NOD.LCMV TCR Tg stock than in standard NOD mice, such effectors might undergo a highly efficient expansion even before their first antigenic encounter.

The flow cytometric data described above indicate that although they might undergo an efficient expansion once generated, the overall precursor frequency of nontransgenic T-cell clonotypes is likely to be less in NOD.LCMV TCR Tg than in standard NOD mice. However, to more definitively address this issue, we determined if the NOD.LCMV TCR Tg stock has a lower frequency than standard NOD mice of CD4 or CD8 T-cells expressing Vβ elements differing from that characterizing the LCMV TCR Tg. Our original analyses depicted in Table 1 indicated that ∼75% of the CD4 T-cells that aberrantly developed in the NOD.LCMV TCR Tg stock failed to express the transgenic TCR. Confirmation was provided by the finding that, compared with standard NOD mice, the proportion of CD4 T-cells using TCR Vβ elements differing from the transgenic TCR (Vβ8) was reduced by ∼25% (Fig. 2A). This finding, combined with the fact they were characterized by approximately fourfold lower numbers of total CD4 T-cells than standard NOD mice, indicates that the precursor frequency of diabetogenic CD4 effectors should be greatly reduced in the NOD.LCMV TCR Tg stock. Functional support for this conclusion was provided by the finding that the CD4 T-cell proliferative response to the 65-kDa isoform of the candidate β-cell autoantigen GAD was ∼25% less in NOD.LCMV TCR Tg than in standard NOD mice (data not shown). Furthermore, compared with standard NOD mice, only a barely detectable proportion of CD8 T-cells in the NOD.LCMV TCR Tg stock used nontransgenic TCR elements (Fig. 2B). Hence, NOD mice expressing the LCMV TCR Tg are indeed characterized by a greatly reduced frequency of other T-cell clonotypes, especially in the CD8 compartment, presumably including those normally contributing to type 1 diabetes.

We considered two explanations for the full type 1 diabetes susceptibility of NOD.LCMV TCR Tg mice. The first was that although their precursor frequency was greatly reduced because of the allelic exclusion induced by the pathologically irrelevant transgenic TCR, an ability to undergo efficient expansion allowed the residual β-cell autoreactive T-cells present in NOD.LCMV TCR Tg mice to mediate full pathogenic effects. Alternatively, it was also possible that type 1 diabetes development in NOD.LCMV TCR Tg mice resulted from their pancreatic β-cells displaying a MHC class I−bound autoantigenic peptide that could be recognized in a cross-reactive fashion by the transgenic TCR. As an initial means of testing these possibilities, we determined if CD4 and CD8 T-cells purified from 6- to 8-week-old standard NOD or NOD.LCMV TCR Tg female mice could alone, or in various combinations, transfer type 1 diabetes to NOD-scid recipients. A previous study (36) demonstrated that CD4 or CD8 T-cells from young prediabetic NOD donors did not individually transfer disease to NOD-scid recipients, but could do so when admixed. We obtained virtually identical results in the current study (Table 2). Similarly, CD4 or CD8 T-cells from young NOD.LCMV TCR Tg donors did not individually transfer type 1 diabetes to any NOD-scid recipients. In contrast, type 1 diabetes developed in 80% (16/20) of NOD-scid mice repopulated with a mixture of CD4 and CD8 T-cells purified from young standard NOD and NOD.LCMV TCR Tg donors, respectively. These results indicate that type 1 diabetes development in NOD.LCMV TCR Tg mice was not independently mediated by the unusually large pool of nontransgenic CD4 T-cells (see Table 1). Furthermore, these results also indicate that, although mostly expressing a putatively β-cell−irrelevant TCR, CD8 T-cells are essential contributors to type 1 diabetes development in NOD.LCMV TCR Tg mice, but their pathogenic activity is dependent on CD4 T-cell helper activities.

The question still remained whether the diabetogenic activity of CD8 T-cells in NOD.LCMV TCR Tg mice resulted from cross-reactivity of the transgenic TCR against an unknown pancreatic β-cell autoantigen. Another possibility was that the diabetogenic CD8 T-cells resided within the residual subpopulation that used endogenously derived TCR molecules and were capable of undergoing efficient expansion. To distinguish between these possibilities, we exploited the scid mutation that blocks productive rearrangement of germ line TCR gene segments (37). Because the scid mutation had already been fixed on the inbred NOD background (27), it was possible to produce NOD-scid.LCMV TCR Tg mice through a single outcross-intercross cycle. FACS analysis revealed a complete absence of CD4 T-cells in these NOD-scid.LCMV TCR Tg mice (Fig. 3). Hence the subpopulation of CD4 T-cells expressing the transgenic TCR that developed in standard NOD.LCMV TCR Tg mice was actually selected through expression of a second endogenously derived TCR. In contrast, CD8 T-cells, which all expressed the transgenic TCR, remained present in NOD-scid.LCMV TCR Tg mice. Most importantly, NOD-scid.LCMV TCR Tg mice remained completely free of type 1 diabetes (Table 3). These results indicate that the transgenic LCMV-reactive CD8 T-cells could not independently mediate type 1 diabetes development in NOD mice by mounting a cross-reactive response against pancreatic β-cells.

Based on the adoptive transfer studies described earlier, one possible explanation for the inability of transgenic LCMV-reactive CD8 T-cells to independently mediate type 1 diabetes development in NOD-scid.LCMV TCR Tg mice was a lack of helper activities provided by CD4 T-cells. Thus, we determined if NOD-scid.LCMV TCR Tg mice developed type 1 diabetes after an infusion of CD4 T-cells purified from standard NOD donors. As shown in Table 3, these NOD-scid.LCMV TCR Tg mice were successfully repopulated with CD4 T-cells, but remained completely free of type 1 diabetes. These results indicate that, even when provided with a source of helper activity, CD8 T-cells expressing the transgenic LCMV-reactive TCR did not exert diabetogenic activity in NOD mice by mounting a cross-reactive response against pancreatic β-cells. Furthermore, these collective results also indicate that the CD8 T-cells that did contribute to type 1 diabetes development in standard NOD.LCMV TCR Tg mice reside within the residual subpopulation that bypasses allelic exclusion, express endogenously derived TCR molecules, and undergo efficient expansion. However, as a further test of this issue, we also evaluated whether CD8 T-cells propagated from the islets of NOD.LCMV TCR Tg mice truly resided within the residual nontransgenic compartment. Based on the absence of TCR Vα2 expression, the vast majority (∼92%) of islet-reactive CD8 T-cells from NOD.LCMV TCR Tg mice did indeed reside in the residual nontransgenic compartment (Fig. 4). Virtually all (∼99%) of β-cell autoreactive CD8 T-cells isolated from standard NOD mice were also TCR Vα2 negative.

An array of β-cell autoreactive effectors broad enough to elicit type 1 diabetes development could clearly be generated from within the overall small proportion of CD8 T-cells that bypassed allelic exclusion in NOD.LCMV TCR Tg mice. However, it was possible that the process of allelic exclusion did result in a lower frequency of certain β-cell autoreactive CD8 T-cell populations in the NOD.LCMV TCR Tg stock than in standard NOD mice. The K^d-bound NRP-A7 peptide represents a mimotope of the β-cell autoantigen recognized by a prevalent population of diabetogenic CD8 T-cells in standard NOD mice (20). A characteristic of the first CD8 T-cell clone (NY8.3) found to recognize K^d/NRP-A7 was a TCR containing a GGSNAKLT motif in the CDR3α region (20). We subsequently isolated another diabetogenic CD8 T-cell clone from NOD mice, AI12.B1.3, that was characterized by the identical CDR3α GGSNAKLT motif but a very different TCR-β chain than that found in the NY8.3 clone (16). As shown in Fig. 5, a K^d/NRP-A7 tetramer was recognized by a lymphoma transfected to express the AI12-B1.3 TCR, but not by the previously described K^d-restricted β-cell autoreactive T-cell clone, AI4, which does not use a CDR3α GGSNAKLT motif. Furthermore, the transfected lymphoma expressing the AI12.B1.3 TCR secreted IL-2 when stimulated with RMA-S/K^d-cells pulsed with NRP-A7 but not with an insulin control peptide, indicating that tetramer binding to these cells was an antigen-specific event (Table 4). Thus, we were able to use the AI12.B1.3 TCR expressing transfectant as a positive control in a tetramer-binding assay that compared the frequency of CD8 T-cells capable of recognizing K^d/NRP-A7 in the islets of NOD.LCMV TCR Tg and standard NOD mice. Significantly, as shown in Fig. 6, the proportion of β-cell autoreactive CD8 T-cells from NOD.LCMV TCR Tg mice that recognized K^d/NRP-A7 (∼22%) approached a level similar to that of CD8 T-cells from standard NOD mice (∼39%). These results indicate that, even under conditions in which their precursor pool is drastically reduced, β-cell autoreactive CD8 T-cells contributing to type 1 diabetes development in NOD mice undergo a highly efficient expansion.

Our results indicate that, even under conditions in which their precursor frequency is greatly reduced through the allelic exclusion induced by a pathologically irrelevant transgenic TCR, β-cell autoreactive T-cells in NOD mice still efficiently expand to levels necessary for the development of overt type 1 diabetes. This was best illustrated by the fact that a β-cell autoreactive CD8 T-cell population that was prevalently observed in standard NOD mice remained present at high levels in pancreatic islets of a NOD stock in which the total T-cell repertoire was predominantly skewed toward recognition of an MHC class I−restricted LCMV peptide. Although nontransgenic CD8 T-cells were essential contributors to the normal rate of type 1 diabetes development in the NOD.LCMV TCR Tg stock, the disease process also required contributions from MHC class II−restricted CD4 T-cells. This indicated that the high rate of allelic exclusion induced by the transgenic MHC class I−restricted, LCMV-specific TCR did not prevent NOD mice from continuing to produce a sufficient array of diabetogenic CD4 T-cells.

The high numbers of β-cell autoreactive T-cells that continued to accumulate in the NOD.LCMV TCR Tg stock appeared to result from a highly efficient expansion of such effectors, rather than their being generated at the same frequency as in standard NOD mice. We base this conclusion on two observations. The first is that because of the high rate of allelic exclusion elicited by the pathologically irrelevant transgenic TCR, all other TCR clonotypes examined were present at much lower frequencies in NOD.LCMV TCR Tg than in standard NOD mice. Second, although characterized by a slightly reduced frequency of TCR transgenic T-cells, much higher total numbers of both these and nontransgenic T-cells accumulated in NOD than in B6,D2 control mice. Because the T-cells that preferentially accumulate in NOD versus B6,D2 mice include the TCR transgenic clonotype that presumably would not have encountered its cognate LCMV peptide, this phenotype appears to be driven in an antigen-independent fashion. By extension, this indicates that the NOD genetic background may preferentially promote a generalized highly efficient antigen-independent expansion of T-cells. Of course, the residual nontransgenic diabetogenic T-cells that remain present in NOD.LCMV TCR Tg mice will retain the capacity to expand upon recognition of their cognate antigen. However, our current findings indicate that diabetogenic T-cells in NOD mice may also be able to undergo an efficient expansion before encountering antigen that may represent an important and heretofore unconsidered aspect of pathogenesis.

The ability of NOD mice to generate a relatively large number of non-TCR transgenic T-cells suggests that the process of allelic exclusion may be less efficient in these mice than in other inbred strains. This possibility is supported by the fact that, in the presence of the same MHC class I−restricted LCMV-specific transgenic TCR, more nontransgenic MHC class II−restricted CD4 T-cells were produced in NOD mice than in nonautoimmune-prone B6,D2 control animals. Similarly, we previously found that NOD mice transgenically expressing the TCR from an MHC class I−restricted β-cell autoreactive T-cell clone also continued to generate a relatively large number of CD4 T-cells that bypassed allelic exclusion, as evidenced by their coexpression of a second endogenously derived TCR (21). Indeed, it has been proposed that the ability to bypass allelic exclusion and express a second TCR is an important pathogenic mechanism by which β-cell autoreactive T-cells in NOD mice escape normal selection constraints (38). However, the higher levels of nontransgenic CD4 T-cells in NOD than in B6,D2 mice carrying the LCMV TCR Tg could also result from the known defective ability of H2-A^g7 MHC class II molecules expressed in the former strain to induce immunologic tolerance (3).

The current studies provided further insights into the antigen recognition characteristics of TCR molecules used by β-cell autoreactive CD8 T-cells in NOD mice. It has been previously demonstrated that a prevalent population of β-cell autoreactive CD8 T-cells in standard NOD mice use a TCR-α chain characterized by a GGSNAKLT motif in the CDR3 loop encoded by a Vα17 to Jα42 gene rearrangement event (15,16). The NRP-A7 peptide presented by H2-K^d represents a mimotope of the antigen recognized by the NOD β-cell autoreactive CD8 T-cell clone NY8.3, which uses a CDR3α GGSNAKLT motif encoded by a Vα17 to Jα42 TCR gene rearrangement (20). Our K^d-restricted β-cell autoreactive AI12.B1.3 T-cell clone was also characterized by a TCR-α chain with a CDR3 GGSNAKLT motif, but its TCR-β chain was very different from that found in the NY8.3 clone (16). Nonetheless, here we showed that, like NY8.3, the AI12.B1.3 clone also recognized K^d/NRP-A7. Coupled with the prevalent usage of such TCR gene rearrangements by diabetogenic CD8 T-cells in NOD mice, these recognition data suggest that TCR-α chains encoded by a Vα17 to Jα42 gene rearrangement may be particularly important in conferring reactivity against the β-cell autoantigen mimicked by the K^d/NRP-A7 complex.

As noted above, the fact that K^d/NRP-A7 reactive CD8 T-cells continued to accumulate at high levels in the islets of NOD.LCMV TCR Tg mice also provided strong evidence that NOD β-cell autoreactive T-cells can undergo a highly efficient expansion from a limited pool of precursors. The highly efficient expansive capacity of such autoreactive effectors also probably accounts for why type 1 diabetes developed at a high rate in a NOD stock transgenically expressing another putatively nonpathogenic TCR (39), rather than the explanation proposed in this earlier study (39) that under some circumstances, T-cells can mediate β-cell destruction in a nonspecific (i.e., antigen-independent) fashion.

In conclusion, our results indicate that the accumulation of β-cell−autoreactive T-cells in NOD mice is a highly efficient process, even under conditions that greatly limit their precursor frequency. Significantly, it appears that such an expansion can potentially occur in an antigen-independent fashion. This potential ability of β-cell autoreactive T-cells to efficiently expand before their first encounter with antigen represents a previously unconsidered and perhaps important aspect of type 1 diabetes development in NOD mice. Hence, it should be taken into account that a similar situation may exist in humans deemed at high risk for type 1 diabetes development when developing possible disease intervention protocols.

This work was supported by National Institutes of Health Grants DK-46266 (D.V.S), DK-51090 (D.V.S), AI-41469 (D.V.S.), AI-28802 (S.G.N.), AI-07289 (S.G.N.), DK-52956 (S.G.N.), HL-54977 (S.J.), F-32-DK-09889 (R.T.G.), and Cancer Center Support Grant CA-34196 (The Jackson Laboratory), as well as by grants from the Juvenile Diabetes Foundation International (D.V.S., S.G.N., and S.J.). T.P.D. is a fellow of the Cancer Research Institute.