Two novel transgenic (Tg) strains were created expressing hen egg-white lysozyme (HEL) in a pancreas-specific fashion. RmHP.111 mice had levels of HEL per cell similar to that of the established ILK-3 strain, while RmHP.117 mice had 10-fold lower levels (50,000 molecules per cell). When bred to 3A9 TCR Tg mice, negative selection occurred equally in all three double-Tg combinations, yet only ILK-3 × 3A9 and RmHP.111 × 3A9 mice became diabetic. Additionally, activated 3A9 cells readily transferred diabetes into ILK-3 or RmHP.111 mice, but only marginally into the RmHP.117 strain. In the peripancreatic lymph node, division of naive 3A9 cells was similar between RmHP.111 and RmHP.117 strains, but pancreatic APCs from RmHP.111 × 3A9 mice stimulated HEL-reactive cells to a much greater degree than those from RmHP.117 × 3A9 mice. In this model, diabetes was dependent upon both initial priming in the peripancreatic lymph node and subsequent presentation in the pancreas, with disease incidence predicted by the β cell level of autoantigen.

Type I diabetes is a complex multivariable disease in which autoreactive lymphocytes cause tissue destruction and functional impairment of the pancreas. Many genes have been implicated in diabetes pathogenesis (1), chief among them are those encoding the class II molecule in humans and mice (Refs.2 and 3 , reviewed in Ref.4). However, despite an accumulating amount of evidence, much remains unknown about the events precipitating the initial autoimmune insult and the secondary events necessary to progress to eventual disease.

To model diabetogenesis, many researchers have used the rat insulin promoter (RIP)3 to drive pancreatic expression of a wide variety of genes, beginning with Hanahan’s initial characterization of the SV-40 large T Ag (5). Some RIP-Ag transgenic (Tg) models developed spontaneous insulitis or became diabetic when crossed with the appropriate TCR Tg mice (6, 7, 8, 9, 10, 11, 12, 13). In other models, the T cells must first be stimulated by infection or immunization (14, 15). Some reports focused exclusively on CD8 responses (14, 16, 17, 18, 19), looked primarily at transfer studies (20, 21, 22), or used the insulin promoter to express soluble and not membrane-bound proteins (17).

Within the class II system, where cross-presentation of β cell Ags from a professional APC is required, a number of parameters influence disease development. Frequency of reactive T cells (8), levels of IFN-γ (6), and the affinity of the Tg TCR (7) were shown to be important factors controlling the autoimmune T cell response. Ag levels might also be a key factor in the initiation or progression of diabetic autoimmunity. In support of this, studies demonstrated that high levels of pancreas-expressed OVA protein led to greater division of Tg CD8 T cells in the peripancreatic lymph node (pLN), whereas lower levels of protein engendered no such division, although diabetes incidence was not addressed in this model (17, 21).

In the 3A9 TCR Tg system (recognizing an epitope of the hen egg-white lysozyme (HEL) protein), it was initially demonstrated that 100% of 3A9 mice crossed to the ILK-3 strain (expressing HEL in a pancreas-specific fashion) exhibited massive insulitis (10). Later studies reported that a variable but reproducible number of ILK-3 × 3A9 double-Tg mice became spontaneously diabetic (12, 13). In all reports, insulitis and/or diabetes occurred despite robust negative selection of the 3A9 T cells (12, 13, 23, 24). It was reasoned that diabetes occurred because a few high avidity 3A9 T cells escaped negative selection, encountered a high level of HEL in the periphery, became activated, and subsequently precipitated disease. However, to date, there is a paucity of evidence in any CD4 model of diabetes on how varying the level of autoantigen per β cell influences initiation or exacerbation of an autoimmune phenotype.

Because Ag may be presented at many points during the lifespan of a T cell, diabetogenesis has been modeled as a process of multiple steps or checkpoints (25). The initial step is selection of T cells in the thymus, followed by migration to the periphery, where T cells encounter Ag in the pLN. If primed correctly, newly activated cells then distribute to the pancreas, instituting conditions from mild insulitis to fulminant diabetes. Many studies have focused on either the thymus as a site of selecting autoimmune T cells (13, 26, 27), or the pLN as the next location controlling disease propagation (28, 29). Final progression to full-blown disease is less well studied, but is becoming an area of active investigation (30).

To determine how the level of autoantigen per β cell affects an ensuing CD4 T cell response, and to quantitate the level of Ag necessary for diabetes induction within a single-Tg TCR system, we created two novel strains of RIP-mHEL mice and subsequently crossed them to the 3A9 TCR Tg strain. In this model, negative selection of 3A9 T cells occurred equally well between RIP-mHEL strains that bore vastly different levels of membrane and serum target protein. After central selection, some Tg T cells escaped to the periphery where the level of Ag per islet β cell became a critical factor in determining the incidence of spontaneous diabetes. In this, the pancreas was found to become an active site of Ag presentation and APC accumulation, with the level of presentation from pancreatic APCs correlating with disease incidence. In total, these data allowed us to construct a model emphasizing the prominent factors underlying progression to full blown diabetes.

B10.BR mice were purchased from The Jackson Laboratory and maintained as a breeding colony in our facility. 3A9 mice were a kind gift from Dr. M. Davis (Stanford University, Stanford, CA), and the ILK-3 strain was obtained from the laboratory of Dr. C. Goodnow (Australian National University, Canberra, Australia). Lines were developed expressing a membrane form of HEL (mHEL) under the rat insulin promoter (RmHP). RmHP Tg mice were generated by microinjecting a RIP-mHEL construct into fertilized oocytes generated from a C57BL/6 × B10.BR mating. Positive mice were first screened by PCR, confirmed by Southern blotting with a 350-bp HEL probe, and subsequently bred to the B10.BR strain a further four generations. The RIP-mHEL construct consisted of 660 bp of the RIP (5) placed upstream of a HEL cDNA sequence, which, in turn, was fused to exon 4 of the genomic sequence of the class I molecule Ld encoding for the transmembrane region (31). Downstream of the entire coding sequence was a single loxP site. RmHP.117 mice were generated by obtaining an original line with multiple integrations of the construct and subsequently lowering the copy number by interbreeding with the Cre-deletor strain (32) on a B10.BR background. None of the single-Tg strains demonstrated any signs of diabetes, and all single-Tg lines appeared viable, fertile, and otherwise healthy in every respect.

Islets of Langerhans were isolated by collagenase digestion (33) to obtain a single suspension of β cells. For β cells, equal numbers were lysed in a protein lysis solution (50 mM Tris (pH 8.0), 250 mM NaCl, 5 mM EDTA, and 0.5% Nonidet P-40 containing the protease inhibitors pepstatin A (2.5 μg/ml), leupeptin (5.0 μg/ml), and PMSF (1 mM)) and an anti-HEL ELISA was conducted by overnight coating of a Nunc-Immuno plate (Fisher Scientific) with 5 μg/ml of the anti-HEL mAb F10.6.6 (34), blocking with PBS/1% BSA, and adding protein lysates. Bound HEL was detected with a purified rabbit anti-HEL Ab, followed by a biotinylated sheep anti-rabbit Ab (Sigma-Aldrich), and finally a streptavidin-HRP conjugate (Sigma-Aldrich), was developed with ABTS (Roche), and was detected at 405 nm. To calculate the molecules of HEL per β cell, multiple points in the linear part of the anti-HEL ELISA curve for each β cell prep were compared directly with readings for a known amount of HEL run in the same assay.

For HEL detection, a biotinylated F10.6.6 Ab was used, followed by streptavidin-PE (Molecular Probes). In some control experiments, polyclonal rabbit anti-HEL Abs (produced by repeated immunization of rabbits with purified HEL) were used. The purified IgG Abs bound to native and denatured HEL (tested via Western blotting). The polyclonal Abs gave a similar staining profile of β cell-associated HEL as the mAb F10.6.6. To determine mean fluorescence intensity (MFI) of anti-HEL β cell staining for each strain, the baseline value of staining B10.BR β cells (7.67) was subtracted from the calculated geometric mean value for each peak.

The 1G12 Ab was previously described (31) and all other Abs, including CD8-FITC, CD25-PE, CD4-PECy7, were purchased from BD Pharmingen. Streptavidin-allophycocyanin was also occasionally used as a secondary reagent (Molecular Probes). In all flow cytometry staining, the 1G12 Ab was used to detect Tg+ 3A9 T cells. For immunofluorescence, pancreata were frozen in OCT and sections were cut at 5–8 μm, blocked with 5% goat serum (Vector Laboratories), and stained. For class II staining, the anti-I-Ak-FITC Ab (BD Pharmingen) was followed by an anti-fluorescein Ab coupled to Alexa-488 (Molecular Probes). Despite numerous attempts, the 1G12 Ab was unable to be used for immunofluorescent staining on frozen tissue sections. As a proxy, Vβ 8.1/8.2+ cells were detected with a biotinylated anti-Vβ 8.1/8.2 mAb (BD Pharmingen) followed by streptavidin-Alexa 555 (Molecular Probes).

For cross-presentation of HEL-bearing islets, 3.3 × 105 unpurified splenocytes from a 3A9 TCR Tg mouse (containing ∼2 × 104 CD4+1G12+ cells) were plated with varying amounts of HEL+ β cells. Supernatant was removed at 24 h to test for IL-2, and cultures were pulsed with [3H]thymidine and harvested at 48 h. IL-2 was measured with a standard CTL line assay as previously described. Mouse β cells and pancreas infiltrating cells were isolated by collagenase digestion (33) and purified over a Ficoll gradient. Single cell suspensions were generated by briefly incubating whole islets with 1 mg/ml trypsin. Dispersed cells were evaluated for either the level of HEL staining in the case of single-Tg mice or for the percentage and total number of class II+ cells in the case of double-Tg mice. From immunofluorescent staining of pancreatic frozen sections, it is clear that class II+ cells encompass both CD19+ and CD19 cell populations. Further flow cytometric analysis was not performed. To evaluate the dose response potential of double-Tg cells, lymph nodes bearing CD4+1G12+ cells from two mice were enriched for T cells using a CD3 enrichment column (R&D Systems), the total number of cells per well was equalized between groups with naive B10.BR splenocytes and plated with varying amounts of HEL. Pulsing and recovery of supernatant was as above. Similar results were seen using enriched splenocytes from double-Tg animals. For pancreas-derived APC presentation, 1 × 105 3A9 hybridoma cells were cultured with varying numbers of irradiated infiltrating APCs, and IL-2 was measured at 24 h.

Naive 3A9 cells from lymph nodes and spleen were labeled with 5 μM CFSE (Molecular Probes) as described elsewhere (35) and transferred into unmanipulated single-Tg mice at a concentration of 1.5 × 106 CD4+1G12+ cells per mouse. pLN or brachial lymph nodes were removed 3 days after transfer and were assayed for division. Plots shown are gated on CD4+1G12+ cells. For transfer of in vitro-activated cells, naive 3A9 lymphocytes from lymph nodes and spleen were stimulated for 3 days ex vivo with 10 μM HEL (Sigma-Aldrich), spun over histopaque, and transferred i.v. into single-Tg mice irradiated the previous day with 600 Gy. Each mouse received 2.5 × 107 total cells, containing ∼1 × 107 CD4+1G12+ cells.

Two novel lines of RIP-mHEL Tg mice were generated, bearing detectable levels of HEL in the pancreas (called hereafter RmHP.111 and RmHP.117). Similar to ILK-3 mice (10), HEL protein was also detected, at considerably lower levels, in both the kidney and mesenteric lymph nodes. No HEL protein was detected in the thymus or a variety of other organs (with an anti-HEL ELISA detectable to 10 fmol) assaying up to 300 μg of tissue (data not shown). For serum levels of HEL, ILK-3 mice had an average concentration of 1 ng/ml, as previously reported, while the serum level in RmHP.111 mice was 20-fold lower at 0.043 ng/ml, and was below detection (<0.01 ng/ml) in the RmHP.117 strain (Fig. 1 A). Importantly, none of the single-Tg mice became diabetic or showed histopathological evidence of insulitis.

FIGURE 1.

HEL quantitation in the serum or per β cell. A, The anti-HEL ELISA determined the level of HEL in the serum of each strain (level of detection to 0.01 ng/ml). B, The mAb F10.6.6 detected the level of membrane-bound HEL per β cell (solid line, ILK-3; dotted line, RmHP.111; dashed line, RmHP.117; solid line on far left, B10.BR ctrl). C, Primary proliferation of naive 3A9 splenocytes cross-presented with collagenase-isolated β cells from each of the three single-Tg strains; D, the indirect measure of IL-2 production for the same assay. E, Anti-HEL ELISA results for the same population of collagenase-isolated β cells (for CE, ▪, ILK-3; ▴, RmHP.111; and ▾, RmHP.117).

FIGURE 1.

HEL quantitation in the serum or per β cell. A, The anti-HEL ELISA determined the level of HEL in the serum of each strain (level of detection to 0.01 ng/ml). B, The mAb F10.6.6 detected the level of membrane-bound HEL per β cell (solid line, ILK-3; dotted line, RmHP.111; dashed line, RmHP.117; solid line on far left, B10.BR ctrl). C, Primary proliferation of naive 3A9 splenocytes cross-presented with collagenase-isolated β cells from each of the three single-Tg strains; D, the indirect measure of IL-2 production for the same assay. E, Anti-HEL ELISA results for the same population of collagenase-isolated β cells (for CE, ▪, ILK-3; ▴, RmHP.111; and ▾, RmHP.117).

Close modal

To quantitate β cell levels of HEL, islets were isolated from each strain, dispersed to single cells, and the individual β cells were stained with the anti-HEL Ab F10.6.6 (34). As is shown in Fig. 1 B, ILK-3 and RmHP.111 β cells exhibited a nearly equivalent amount of HEL on the surface of the β cell (MFI minus background staining of 2253 and 1792, respectively), while staining of RmHP.117 β cells was ∼7-fold less with a MFI of 259. Flow cytometry using FITC-conjugated rabbit anti-HEL polyclonal Abs gave a similar pattern of staining between the three transgenics (data not shown).

To assay the amount of HEL in a functional assay, β cells from single-Tg mice were cross-presented to HEL-reactive naive 3A9 splenocytes, and the T cell response was measured by both primary proliferation and production of IL-2. For primary proliferation, 1–2 × 103 β cells from either ILK-3 or RmHP.111 single-Tg mice stimulated 3A9 T cells to a half-maximal response (Fig. 1,C). In contrast, it took >1 × 105 β cells from the RmHP.117 strain to achieve a similar level of primary stimulation. Results with IL-2 production varied only slightly (Fig. 1 D). Here, β cells from ILK-3 mice were ∼4-fold more efficient at generating IL-2 in the culture than RmHP.111 islet cells. Again, RmHP.117 cells were extremely inefficient by comparison, requiring >3 × 105 β cells to generate a similar level of IL-2. There was no proliferation or IL-2 production above background when non-Tg β cells were used as a source of Ag.

Lastly, an anti-HEL ELISA was performed on a purified population of β cells from each single-Tg strain. Consistent with the functional data, ILK-3 and RmHP.111 mice demonstrated nearly equivalent amounts of HEL per β cell by ELISA. The amount of HEL per β cell for the RmHP.117 strain was lower by >10-fold. By direct comparison to a HEL standard, each β cell from ILK-3, RmHP.111, and RmHP.117 mice contained 26.8, 17.9, and 1.14 fg of HEL, respectively. These amounts translated to 1.12 × 106, 7.45 × 105, and 4.7 × 104 molecules of HEL per β cell in the ILK-3, RmHP.111, and RmHP.117 strains. A summary of HEL quantitation is displayed in Table I.

Table I.

Quantitation of HEL

ILK-3RmHP.111RmHP.117
Serum HEL (ng/ml) 1.0 0.043 ND (<0.01) 
F10 staining (MFI) 2253 1784 259 
HEL ELISA    
 Femtograms per/cell 26.8 17.9 1.14 
 Molecules per cell 1.12 × 106 7.45 × 105 4.7 × 104 
Cross-presentation (1/2 max)    
 Primary 1 × 103 2 × 103 2 × 105 
 IL-2 (CTLL) 1 × 103 4 × 103 >3 × 105 
ILK-3RmHP.111RmHP.117
Serum HEL (ng/ml) 1.0 0.043 ND (<0.01) 
F10 staining (MFI) 2253 1784 259 
HEL ELISA    
 Femtograms per/cell 26.8 17.9 1.14 
 Molecules per cell 1.12 × 106 7.45 × 105 4.7 × 104 
Cross-presentation (1/2 max)    
 Primary 1 × 103 2 × 103 2 × 105 
 IL-2 (CTLL) 1 × 103 4 × 103 >3 × 105 

To assay for diabetes susceptibility, ILK-3, RmHP.111, or RmHP.117 mice were intercrossed with the 3A9 TCR Tg strain, and double-Tg mice were monitored on a weekly basis for elevations in serum glucose. As expected, ILK-3 × 3A9 mice became diabetic at the predicted rate, with incidence reaching nearly 60% by 20 wk of age (Fig. 2,A). RmHP.111 × 3A9 mice also became diabetic, with a penetrance and incidence equivalent or slightly greater than that of ILK-3 × 3A9. None of the 15 RmHP.117 × 3A9 mice, followed for >20 wk, ever became diabetic. In both the ILK-3 × 3A9 and RmHP.111 × 3A9 strains, diabetes was characterized by lymphocyte and large mononuclear infiltrates surrounding, and within, the islets (Fig. 2,B). Although RmHP.117 × 3A9 mice never became diabetic, mononuclear infiltrates were present in some islets: 60% of islets had no detectable infiltration (Fig. 2,C, top), another 30% contained a small degree of peri-insulitis, and the remaining 10% contained a pronounced infiltration that occupied greater than half of the total islet volume (Fig. 2 C, bottom).

FIGURE 2.

Diabetes incidence and negative selection in RIP-mHELx 3A9 mice. Diabetes incidence (two serum glucose readings >250 mg/dl) for each of the RIP-mHELx3A9 combinations is shown in A (▪, ILK-3 × 3A9 (n = 22); ▴, RmHP.111 × 3A9 (n = 16); ▾, RmHP.117 × 3A9 (n = 15)). B, An infiltrated islet typical of diabetic RmHP.111 × 3A9 animals at 5 wk of age; C, an example of both an unaffected islet (60% of total islets) and a heavily infiltrated islet (10% of total) from a RmHP.117 × 3A9 mouse. D, CD4 vs CD8 profiles of thymocytes gated by forward and side scatter are shown (upper left, 3A9; upper right, ILK-3 × 3A9; lower left, RmHP.117 × 3A9; lower right, RmHP.111 × 3A9). E, Shown are the total number of thymocytes (upper panel) or percentage of peripheral lymphocytes (lower panel) that are CD4+1G12+CD8.

FIGURE 2.

Diabetes incidence and negative selection in RIP-mHELx 3A9 mice. Diabetes incidence (two serum glucose readings >250 mg/dl) for each of the RIP-mHELx3A9 combinations is shown in A (▪, ILK-3 × 3A9 (n = 22); ▴, RmHP.111 × 3A9 (n = 16); ▾, RmHP.117 × 3A9 (n = 15)). B, An infiltrated islet typical of diabetic RmHP.111 × 3A9 animals at 5 wk of age; C, an example of both an unaffected islet (60% of total islets) and a heavily infiltrated islet (10% of total) from a RmHP.117 × 3A9 mouse. D, CD4 vs CD8 profiles of thymocytes gated by forward and side scatter are shown (upper left, 3A9; upper right, ILK-3 × 3A9; lower left, RmHP.117 × 3A9; lower right, RmHP.111 × 3A9). E, Shown are the total number of thymocytes (upper panel) or percentage of peripheral lymphocytes (lower panel) that are CD4+1G12+CD8.

Close modal

Thymic selection of 3A9 T cells was then evaluated in each of the double-Tg crosses. Representative thymus plots are shown in Fig. 2,D (CD4 vs CD8 profiles of forward and side-scatter gated thymocytes) with graphic representation in Fig. 2,E and a summary in Table II. For each double-Tg combination, thymic selection looked remarkably similar. In 3A9 single-Tg animals, nearly 30% of thymocytes were CD4 single positive (SP; Fig. 2,D, upper left) and >90% of these were CD4 SP, 1G12+. In contrast, CD4 SP percentages dropped to 4–6% of total thymocytes (Fig. 2,D), with only 50–65% of CD4 SP cells bearing the 1G12 clonotype in any double-Tg combination. In terms of total numbers, there were equivalent total numbers of CD4 SP and CD4 SP, 1G12+ in the thymi of each double-Tg cross compared with each other (e.g., 1.19 ± 0.85 vs 0.91 ± 0.12), but, in all cases, these levels were much lower than those found in 3A9 single-Tg mice (Fig. 2 E).

Table II.

Thymic and peripheral profiles in RIP-HEL×3A9 mice

3A9ILK-3×3A9111×3A9117×3A9
Thymus     
 Total no. (×10−653.0 ± 28.57 21.5 ± 15.41 23.6 ± 3.11 27.51 ± 12.86 
 Percent CD4 SP 29.2 ± 2.69 6.73 ± 0.24 4.58 ± 0.03 4.91 ± 2.77 
 Total no. CD4 SP (×10−614.2 ± 6.60 1.19 ± 0.85 0.91 ± 0.12 1.29 ± 1.20 
 Percentage of CD4 SP that are 1G12+ 91.9 ± 5.58 69.7 ± 11.74 56.0 ± 1.20 63.0 ± 16.19 
 Total no. CD4 SP, 1G12+ (×10−613.17 ± 6.84 0.88 ± 0.73 0.51 ± 0.05 0.91 ± 0.96 
 Percent CD25+× 1G12+ 0.94 ± 0.22 19.0 ± 0.28 33.2 ± 1.98 37.0 ± 5.30 
Spleen     
 Percent CD4+ 6.20 ± 2.93 1.00 ± 0.08 1.45 ± 0.15 1.10 ± 0.07 
 Total no. CD4+ (×10−66.88 ± 3.11 1.93 ± 1.05 1.66 ± 0.09 1.39 ± 0.65 
 Percent CD4+× 1G12+ 4.96 ± 2.96 0.14 ± 0.00 0.17 ± 0.02 0.20 ± 0.04 
 Total no. CD4+, 1G12+ (×10−65.51 ± 3.17 0.26 ± 0.13 0.20 ± 0.01 0.25 ± 0.15 
LN     
 Percent CD4+ 23.35 ± 8.13 4.29 ± 0.62 4.16 ± 0.02 3.64 ± 0.11 
 Percent CD4+1G12+ 22.75 ± 7.56 1.04 ± 0.23 1.02 ± 0.05 0.82 ± 0.47 
3A9ILK-3×3A9111×3A9117×3A9
Thymus     
 Total no. (×10−653.0 ± 28.57 21.5 ± 15.41 23.6 ± 3.11 27.51 ± 12.86 
 Percent CD4 SP 29.2 ± 2.69 6.73 ± 0.24 4.58 ± 0.03 4.91 ± 2.77 
 Total no. CD4 SP (×10−614.2 ± 6.60 1.19 ± 0.85 0.91 ± 0.12 1.29 ± 1.20 
 Percentage of CD4 SP that are 1G12+ 91.9 ± 5.58 69.7 ± 11.74 56.0 ± 1.20 63.0 ± 16.19 
 Total no. CD4 SP, 1G12+ (×10−613.17 ± 6.84 0.88 ± 0.73 0.51 ± 0.05 0.91 ± 0.96 
 Percent CD25+× 1G12+ 0.94 ± 0.22 19.0 ± 0.28 33.2 ± 1.98 37.0 ± 5.30 
Spleen     
 Percent CD4+ 6.20 ± 2.93 1.00 ± 0.08 1.45 ± 0.15 1.10 ± 0.07 
 Total no. CD4+ (×10−66.88 ± 3.11 1.93 ± 1.05 1.66 ± 0.09 1.39 ± 0.65 
 Percent CD4+× 1G12+ 4.96 ± 2.96 0.14 ± 0.00 0.17 ± 0.02 0.20 ± 0.04 
 Total no. CD4+, 1G12+ (×10−65.51 ± 3.17 0.26 ± 0.13 0.20 ± 0.01 0.25 ± 0.15 
LN     
 Percent CD4+ 23.35 ± 8.13 4.29 ± 0.62 4.16 ± 0.02 3.64 ± 0.11 
 Percent CD4+1G12+ 22.75 ± 7.56 1.04 ± 0.23 1.02 ± 0.05 0.82 ± 0.47 

A reduction in clonotype-positive cells was also found in the periphery. Lymph nodes of 3A9 mice contained 23% CD4+ T cells, a great majority of which were CD4+1G12+ (22.75% of total LN cells). In contrast, the levels of CD4+ cells in double-Tg mice approximated 3–4% of total lymphocytes with even fewer (∼1%) being both CD4+1G12+. The spleen gave a similar profile and all staining represents analyses done on three or more separate occasions. Profound negative selection was corroborated by a lack of CD4+1G12+ cells observed upon analysis of peripheral blood (data not shown).

Finally, we also evaluated CD25 expression on clonotype-positive thymocytes. Although 3A9 single-Tg thymocytes expressed CD25 on 1% of CD4 SP, 1G12+ cells, this percentage was much greater in all three double-Tg combinations. In ILK-3 × 3A9 mice, the percentage of CD25+ cells was 19%, while the percentages from RmHP.111 × 3A9 and RmHP.117 × 3A9 were 33 and 37%, respectively. Peripheral activation of 3A9 T cells in double-Tg mice was also assayed via up-regulation of CD69. In the pLN, the percentage of CD69+ staining (on CD4+1G12+ cells) increased from a background of 5% in 3A9 single-Tg mice to >20% in each of the double-Tg crosses and was equivalent between all three double-Tg groups.

To test the ability of cells from double-Tg mice to respond to HEL, spleen and LN cells from double-Tg animals were placed into an in vitro presentation assay with varying amounts of HEL as described in Materials and Methods. As is seen in Fig. 3,A, primary proliferation to HEL was equivalent among lymph node cells from all three double-Tg strains. This was mirrored nearly exactly by their production of IL-2 (Fig. 3 B). Splenocytes from all three strains gave similar results (data not shown).

FIGURE 3.

In vivo proliferation of naive 3A9 T cells in the pLN. A, The primary proliferation (left) or CTL line response (right) of double-Tg cells to graded amounts of HEL (▪, ILK-3 × 3A9; ▴, RmHP.111 × 3A9; ▾, RmHP.117 × 3A9). B, The in vivo response of CFSE-labeled naive 3A9 recovered from the pLN of RmHP.111 (upper left), ILK-3 (upper right), or RmHP.117 (lower left) single-Tg mice. Plots shown are gated on CD4+1G12+ cells and statistics from multiple animals are shown in the bottom right.

FIGURE 3.

In vivo proliferation of naive 3A9 T cells in the pLN. A, The primary proliferation (left) or CTL line response (right) of double-Tg cells to graded amounts of HEL (▪, ILK-3 × 3A9; ▴, RmHP.111 × 3A9; ▾, RmHP.117 × 3A9). B, The in vivo response of CFSE-labeled naive 3A9 recovered from the pLN of RmHP.111 (upper left), ILK-3 (upper right), or RmHP.117 (lower left) single-Tg mice. Plots shown are gated on CD4+1G12+ cells and statistics from multiple animals are shown in the bottom right.

Close modal

In nonobese diabetic (NOD) and some RIP-Tg models, initial presentation to diabetogenic T cells occurred primarily in the pLN (16, 28). To understand primary presentation events in the two novel RIP-mHEL transgenics, naive 3A9 T cells from lymph nodes and spleen were labeled with CFSE and transferred i.v. into each HEL single-Tg strain. Division of 3A9 cells was quantitated 3 days after transfer. CD4+1G12+ cells transferred into ILK-3 mice proliferated in the pLN at a rapid rate, with cells undergoing up to six divisions within the first 72 h (Fig. 3,B, upper right). In contrast, CD4+1G12+ cells in RmHP.111 or RmHP.117 mice did not divide as robustly (Fig. 3 B, upper vs lower left, respectively). Furthermore, despite a large difference in the amount of HEL per β cell, pLN division of 3A9 cells was similar in both RmHP.111 and RmHP.117 recipients.

We next evaluated Ag presentation outside the pLN by looking for class II+ cells in the pancreata of double-Tg mice. Staining for I-Ak revealed a dense interdigitating network of class II-positive cells (green cells in Fig. 4, A and B) in both RmHP.111 × 3A9 and RmHP.117 × 3A9 mice. Interspersed within this APC network were large numbers of Vβ 8.1/8.2-positive cells detected using an anti-Vβ 8.1/8.2 mAb (red cells in Fig. 4, A and B). Single-Tg or B10.BR wild-type pancreata exhibited only sparse and diffuse class II staining and no detectable Vβ 8.1/8.2+ cells (data not shown). Concomitant staining with an anti-HEL mAb easily detected islets bearing HEL in the pancreata of RmHP.117 × 3A9 mice, often in close approximation with class II+ cells. In RmHP.111 × 3A9 pancreata, by contrast, it was more difficult to detect HEL+ or insulin-positive islets, and when detected, the amount of islet remaining was well below that found in unmanipulated single-Tg animals (data not shown).

FIGURE 4.

Visual and functional characterization of APCs from the pancreata of double-Tg mice. A, Frozen pancreas sections from RmHP.111 × 3A9 (left) or RmHP.117 × 3A9 (right) were stained for the presence of I-Ak-positive (green) or Vβ 8.1/8.2-positive (red) cells. Infiltrating APCs from RmHP.111 × 3A9 or RmHP.117 × 3A9 pancreata were recovered (B) and assayed for I-Ak expression by flow cytometry. Shown is the average total number of I-Ak-positive cells recovered per pancreas. C, Titrated numbers of irradiated pancreatic APCs (▴, RmHP.111 × 3A9; ▾, RmHP.117 × 3A9) were incubated with the 48–62 reactive 3A9 hybridoma to test for functional presentation of pancreas-derived HEL.

FIGURE 4.

Visual and functional characterization of APCs from the pancreata of double-Tg mice. A, Frozen pancreas sections from RmHP.111 × 3A9 (left) or RmHP.117 × 3A9 (right) were stained for the presence of I-Ak-positive (green) or Vβ 8.1/8.2-positive (red) cells. Infiltrating APCs from RmHP.111 × 3A9 or RmHP.117 × 3A9 pancreata were recovered (B) and assayed for I-Ak expression by flow cytometry. Shown is the average total number of I-Ak-positive cells recovered per pancreas. C, Titrated numbers of irradiated pancreatic APCs (▴, RmHP.111 × 3A9; ▾, RmHP.117 × 3A9) were incubated with the 48–62 reactive 3A9 hybridoma to test for functional presentation of pancreas-derived HEL.

Close modal

Realizing the pancreas might be a functional site of Ag presentation, the total number and percentage of class II+ cells from islets of normoglycemic RmHP.111 × 3A9 or RmHP.117 × 3A9 mice was quantitated. On average, RmHP.111 × 3A9 mice contained more infiltrating class II+ cells per pancreas than RmHP.117 × 3A9 mice (Fig. 4,B, compare 4.47 × 104 with 2.13 × 104). In addition, when the presentation capacity of pancreas-derived APCs was tested, pancreas-infiltrating cells from RmHP.111 × 3A9 mice stimulated the 3A9 hybridoma to a higher maximal response and triggered IL-2 production at a lower number of class II+ cells per well than did infiltrating APCs from RmHP.117 × 3A9 mice (compare curves in Fig. 4 C).

To test the capacity of each RIP-mHEL host to facilitate disease induction, activated 3A9 cells were transferred into each single-Tg strain and recipients were monitored for diabetes. ILK-3 mice that received in vitro-activated 3A9 cells became diabetic as early as 7 days posttransfer (Fig. 5), with 100% of mice diabetic by day 20. RmHP.111 mice also became diabetic at a rapid rate, the first incidence of hyperglycemia occurring on day 10. In contrast, RmHP.117 mice exhibited a much slower onset of diabetic pathology, taking until day 34 for the first mouse to become diabetic. Furthermore, only 60% of RmHP.117 recipients ever developed diabetes over the 8-wk course of study.

FIGURE 5.

Transfer of diabetes with activated 3A9 T cells. Naive cells from 3A9 TCR Tg mice were activated for 3 days ex vivo with 10 μM HEL and transferred i.v. into each single-Tg strain (▪, ILK-3; ▴, RmHP.111; ▾, RmHP.117). Diabetes was assessed as two consecutive serum glucose readings >250 mg/dl and glucose-positive urine.

FIGURE 5.

Transfer of diabetes with activated 3A9 T cells. Naive cells from 3A9 TCR Tg mice were activated for 3 days ex vivo with 10 μM HEL and transferred i.v. into each single-Tg strain (▪, ILK-3; ▴, RmHP.111; ▾, RmHP.117). Diabetes was assessed as two consecutive serum glucose readings >250 mg/dl and glucose-positive urine.

Close modal

As RmHP.117 × 3A9 mice never became diabetic, the possibility remained that a lack of diabetes was due solely to the influence of regulatory T cells. This was investigated in two ways. First, lymph node cells from RmHP.117 × 3A9 mice containing both CD25+ and CD25 cells were transferred into ILK-3 mice on a Rag−/− background. In this case, two of four recipient mice (50%) became diabetic within 4 wk of transfer. To further rule out the influence of regulatory cells, 5-wk-old RmHP.117 × 3A9 mice (n = 4) were treated with weekly 500-μg injections of anti-CD25 mAb for 4 wk and subsequently followed for the onset of hyperglycemia. Up to 10 wk past the initiation of Ab treatment, none of the mice had become diabetic.

This study evaluated the influence of Ag level on spontaneous diabetes incidence in RIP-mHELx3A9 TCR Tg mice. Having two additional lines of HEL Tg mice, outside of the original ILK-3 strain (10), allowed us to consider various aspects in the initiation and progression of a spontaneous autoimmune response. RmHP.111 mice had only slightly lower levels of HEL per β cell than ILK-3 mice, developed similar diabetes in crosses to the 3A9 strain, and became diabetic after adoptive transfer of activated 3A9 cells in a fashion similar to that seen in ILK-3 mice. The major difference came with the RmHP.117 strain. This strain had 10-fold lower levels of HEL per β cell and was still capable of causing division of naive 3A9 cells in the pLN, but RmHP.117 × 3A9 mice exhibited only mild pathology in the pancreas. Several important findings emerge from these experiments.

First, it was striking that autoimmunity developed in this model despite robust negative selection of 3A9 T cells in the thymus. Indeed, negative selection was similar in all three RIP-mHELx3A9 Tg strains. The robust negative selection was unlikely to be solely a consequence of circulating levels of HEL, as serum levels varied widely between strains, and in the case of RmHP.117 were undetectable by ELISA. It is probable that HEL, by virtue of the RIP, is expressed by specialized thymic epithelial cells, as has been shown for other proteins (36, 37, 38). Thymic HEL presentation might be further influenced by other proteins, including the autoimmune regulator (AIRE), expressed by the thymic medullary epithelium (23, 39).

Despite strong selection pressure in the thymus, 3A9 T cells escaped to the periphery in all three double-Tg strains, were reactive to HEL ex vivo, and caused either diabetes or mild pathology in a fashion correlating with the level of HEL per β cell. To note, despite high percentages of CD4+1G12+CD25+ cells, disease continued to occur in both RmHP111 × 3A9 and ILK-3 × 3A9 mice, suggesting that any obvious control by regulatory cells in these animals was clearly superseded by the response to higher levels of HEL. In RmHP.117 × 3A9 mice, where spontaneous diabetes did not result, regulation was ruled out by either treatment with anti-CD25 Ab (which led to no disease) or the transfer of RmHP.117 × 3A9 cells to ILK-3/Rag−/− mice (with higher levels of HEL per β cell), where disease resulted in a reproducible fashion, despite transfer of CD25+ cells from RmHP.117 × 3A9 mice. This is not to say that CD25+ or Foxp3+ cells play no role in the current model, and indeed, Treg cells continue to be an area of ongoing investigation. However, our data points more prominently toward other influences affecting diabetes progression.

What, then, are other factors that make β cells so sensitive to autoimmune attack in these models? The first may be that the level of autoreactive T cells in any TCR Tg system, relative to a naive primary repertoire, is unusually high. Therefore, despite heavy negative selection, enough T cells are driven to peripheralize and became effector cells. In situations without a Tg TCR (i.e., ILK-3 or RmHP.111 single-Tg animals), diabetes never developed spontaneously. In this case, thymic selection may have been efficient enough to clear away the minimal reactive T cell repertoire before its exit to the periphery. Alternatively, the few cells that could escape in non-TCR Tg systems may be more influenced by peripheral regulatory mechanisms (40). Whether a situation akin to the RIP-mHELx3A9 system exists in non-Tg models of spontaneous disease, like those of NOD, remains unclear. Certainly, there seems to be a predisposition in the NOD strain for less efficient negative selection (13, 24), which then leads to a large export of autoreactive T cells to the periphery (41).

Other features that facilitate initiation of β cell autoimmunity relate to Ag presentation, including efficiency of cross presentation, initial priming of the autoreactive T cell, level of Ag that the T cell subsequently encounters, and continuation of an effector phase. We have shown in previous studies that presentation of β cell-derived HEL was 50-fold more efficient than presentation of soluble HEL (12). In the current model, this presentation gave the APCs of the pLN enough HEL from homeostatic processing to prime a fraction of the 3A9 T cells that escaped negative selection. Indeed, our CFSE results indicated that naive 3A9 cells encountered enough HEL in the pLN of all three strains to undergo some, albeit limited, division.

Interestingly, we did not find a direct relationship between division of naive 3A9 cells in the pLN of single-Tg animals and the amount of HEL per β cell. Stimulation appeared stronger in ILK-3 mice than RmHP.111 mice, which in turn had a response similar to that seen with RmHP.117. Yet RmHP.117 had 10-fold less Ag per β cell, while ILK-3/RmHP.111 mice had nearly similar amounts. Conceivably, ILK-3 mice could have some higher turnover of β cells (resulting in more HEL draining to the node), higher serum levels of HEL in ILK-3 mice could influence pLN division, or other unknown mechanisms could be involved in the transfer of cell-bound HEL to the pLN that amplified the response in ILK-3 mice.

In the β cell, the relation of disease induction to Ag level per cell can be inferred from the RmHP.117 × 3A9 system. Here a lack of diabetes in double-Tg mice establishes a diabetogenic limit of ∼50,000 molecules of HEL per β cell. At this amount of HEL, very mild pathology developed spontaneously in double-Tg mice. Furthermore, transfer of activated 3A9 cells induced a late diabetic pathology in a limited percentage of recipients. It is likely that the weak diabetogenic potential in RmHP.117 × 3A9 mice was secondary to the limited amount of HEL per β cell, a notion substantiated by the ability of RmHP.117 × 3A9 T cells to transfer disease into ILK-3/Rag−/− recipients. As previous studies have shown that a thousand molecules of HEL offered to a single APC resulted in a single 48–62 peptide/MHC complex, a rough estimate can be made that for every β cell taken up by a single APC in RmHP.117 mice, roughly 50 peptide/MHC complexes are generated, a level in this system commensurate with mild pathology, but that does not translate to diabetes.

A final striking feature of our model was the development, in the pancreas, of a large interdigitating network of APCs. In previous studies, mice that lacked pLNs did not develop disease due to a lack of initial priming (42, 43). However, in these studies, when T cells were activated before transfer, even in the absence of peripheral lymph nodes, diabetes was induced. This indicated that further Ag presentation was taking place outside of the lymph node and that this phenomenon could facilitate the effector phase of a diabetic response. Intraislet APC had also previously been shown to be important in both allogenic reactions and to bear β cell Ags (44). This evidence suggested that pancreatic Ag presentation is an important factor outside of the initial priming in the pLN.

These previous results are extended in the current study by recovering APCs from the pancreata of normoglycemic double-Tg mice and demonstrating their Ag presenting capacity. In theory, presentation in the pancreas could amplify an ongoing response by presenting Ag where it is found in highest abundance, which, in turn, might be a key factor in transition from a state of mild insulitis to that of full blown disease. In the current model, it could be envisioned that T cells escaping from negative selection similarly in RmHP.111 × 3A9 and RmHP.117 × 3A9 are stimulated to a similar level in the pLN and then migrate to the pancreas. Here they encounter Ag presented by intraislet APCs, where the level of presentation reflects the local level of HEL per β cell (RmHP.111 ≫ RmHP.117). Inflammation and β cell damage would then ensue.

Theoretically, every time a β cell is destroyed in RmHP.111 mice this would lead (relative to RmHP.117) to 10-fold more Ag donated to the system. This would cause more T cells to divide, a greater recruitment of accessory cells (including APCs) to the pancreas (seen in our study as more class II+ cells/pancreas), and destruction of more β cells. The whole process would then become a feed-forward process, related to the level of Ag per β cell, where more Ag would invariably lead to greater numbers of APCs bearing higher levels of peptide/MHC complexes. Indeed, this scenario is consistent with our analysis of pancreas-derived APCs from RmHP.111 × 3A9 mice. It is interesting to speculate, then, that although not capable of initiating T cell priming, pancreatic APC presentation nevertheless might represent an important step in the amplification of an ongoing response, balancing between a state of mild insulitis and one of full blown disease.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by grants from the National Institutes of Health and the Kilo Diabetes and Vascular Research Foundation.

3

Abbreviations used in this paper: RIP, rat insulin promoter; HEL, hen egg-white lysozyme; pLN, peripancreatic lymph node; Tg, transgenic; MFI, mean fluorescence intensity; SP, single positive; NOD, nonobese diabetic; mHEL, membrane form of HEL.

1
Wicker, L. S., J. A. Todd, L. B. Peterson.
1995
. Genetic control of autoimmune diabetes in the NOD mouse.
Annu. Rev. Immunol.
13
:
179
.-200.
2
Todd, J. A., J. I. Bell, H. O. McDevitt.
1987
. HLA-DQ β gene contributes to susceptibility and resistance to insulin-dependent diabetes mellitus.
Nature
329
:
599
.-604.
3
Tisch, R., H. McDevitt.
1996
. Insulin-dependent diabetes mellitus.
Cell
85
:
291
.-297.
4
McDevitt, H. O..
1998
. The role of MHC class II molecules in susceptibility and resistance to autoimmunity.
Curr. Opin. Immunol.
10
:
677
.-681.
5
Hanahan, D..
1985
. Heritable formation of pancreatic β-cell tumours in transgenic mice expressing recombinant insulin/simian virus 40 oncogenes.
Nature
315
:
115
.-122.
6
Scott, B., R. Liblau, S. Degermann, L. A. Marconi, L. Ogata, A. J. Caton, H. O. McDevitt, D. Lo.
1994
. A role for non-MHC genetic polymorphism in susceptibility to spontaneous autoimmunity.
Immunity
1
:
73
.-83.
7
Degermann, S., C. Reilly, B. Scott, L. Ogata, H. von Boehmer, D. Lo.
1994
. On the various manifestations of spontaneous autoimmune diabetes in rodent models.
Eur. J. Immunol.
24
:
3155
.-3160.
8
Forster, I., R. Hirose, J. M. Arbeit, B. E. Clausen, D. Hanahan.
1995
. Limited capacity for tolerization of CD4+ T cells specific for a pancreatic β cell neo-antigen.
Immunity
2
:
573
.-585.
9
Morgan, D. J., R. Liblau, B. Scott, S. Fleck, H. O. McDevitt, N. Sarvetnick, D. Lo, L. A. Sherman.
1996
. CD8+ T cell-mediated spontaneous diabetes in neonatal mice.
J. Immunol.
157
:
978
.-983.
10
Akkaraju, S., W. Y. Ho, D. Leong, K. Canaan, M. M. Davis, C. C. Goodnow.
1997
. A range of CD4 T cell tolerance: partial inactivation to organ-specific antigen allows nondestructive thyroiditis or insulitis.
Immunity
7
:
255
.-271.
11
Sarukhan, A., A. Lanoue, A. Franzke, N. Brousse, J. Buer, H. von Boehmer.
1998
. Changes in function of antigen-specific lymphocytes correlating with progression towards diabetes in a transgenic model.
EMBO J.
17
:
71
.-80.
12
DiPaolo, R. J., E. R. Unanue.
2001
. The level of peptide-MHC complex determines the susceptibility to autoimmune diabetes: studies in HEL transgenic mice.
Eur. J. Immunol.
31
:
3453
.-3459.
13
Lesage, S., S. B. Hartley, S. Akkaraju, J. Wilson, M. Townsend, C. C. Goodnow.
2002
. Failure to censor forbidden clones of CD4 T cells in autoimmune diabetes.
J. Exp. Med.
196
:
1175
.-1188.
14
Ohashi, P. S., S. Oehen, K. Buerki, H. Pircher, C. T. Ohashi, B. Odermatt, B. Malissen, R. M. Zinkernagel, H. Hengartner.
1991
. Ablation of “tolerance” and induction of diabetes by virus infection in viral antigen transgenic mice.
Cell
65
:
305
.-317.
15
Oldstone, M. B., M. Nerenberg, P. Southern, J. Price, H. Lewicki.
1991
. Virus infection triggers insulin-dependent diabetes mellitus in a transgenic model: role of anti-self (virus) immune response.
Cell
65
:
319
.-331.
16
Kurts, C., H. Kosaka, F. R. Carbone, J. F. Miller, W. R. Heath.
1997
. Class I-restricted cross-presentation of exogenous self-antigens leads to deletion of autoreactive CD8+ T cells.
J. Exp. Med.
186
:
239
.-245.
17
Kurts, C., J. F. Miller, R. M. Subramaniam, F. R. Carbone, W. R. Heath.
1998
. Major histocompatibility complex class I-restricted cross-presentation is biased towards high dose antigens and those released during cellular destruction.
J. Exp. Med.
188
:
409
.-414.
18
Heath, W. R., J. Allison, M. W. Hoffmann, G. Schonrich, G. Hammerling, B. Arnold, J. F. Miller.
1992
. Autoimmune diabetes as a consequence of locally produced interleukin-2.
Nature
359
:
547
.-549.
19
Heath, W. R., C. Kurts, J. F. Miller, F. R. Carbone.
1998
. Cross-tolerance: a pathway for inducing tolerance to peripheral tissue antigens.
J. Exp. Med.
187
:
1549
.-1553.
20
Kurts, C., W. R. Heath, F. R. Carbone, J. Allison, J. F. Miller, H. Kosaka.
1996
. Constitutive class I-restricted exogenous presentation of self antigens in vivo.
J. Exp. Med.
184
:
923
.-930.
21
Li, M., G. M. Davey, R. M. Sutherland, C. Kurts, A. M. Lew, C. Hirst, F. R. Carbone, W. R. Heath.
2001
. Cell-associated ovalbumin is cross-presented much more efficiently than soluble ovalbumin in vivo.
J. Immunol.
166
:
6099
.-6103.
22
Kurts, C., R. M. Sutherland, G. Davey, M. Li, A. M. Lew, E. Blanas, F. R. Carbone, J. F. Miller, W. R. Heath.
1999
. CD8 T cell ignorance or tolerance to islet antigens depends on antigen dose.
Proc. Natl. Acad. Sci. USA
96
:
12703
.-12707.
23
Liston, A., S. Lesage, J. Wilson, L. Peltonen, C. C. Goodnow.
2003
. Aire regulates negative selection of organ-specific T cells.
Nat. Immunol.
4
:
350
.-354.
24
Liston, A., S. Lesage, D. H. Gray, L. A. O’Reilly, A. Strasser, A. M. Fahrer, R. L. Boyd, J. Wilson, A. G. Baxter, E. M. Gallo, et al
2004
. Generalized resistance to thymic deletion in the NOD mouse; a polygenic trait characterized by defective induction of Bim.
Immunity
21
:
817
.-830.
25
Andre, I., A. Gonzalez, B. Wang, J. Katz, C. Benoist, D. Mathis.
1996
. Checkpoints in the progression of autoimmune disease: lessons from diabetes models.
Proc. Natl. Acad. Sci. USA
93
:
2260
.-2263.
26
Mathis, D., C. Benoist.
2004
. Back to central tolerance.
Immunity
20
:
509
.-516.
27
Venanzi, E. S., C. Benoist, D. Mathis.
2004
. Good riddance: thymocyte clonal deletion prevents autoimmunity.
Curr. Opin. Immunol.
16
:
197
.-202.
28
Hoglund, P., J. Mintern, C. Waltzinger, W. Heath, C. Benoist, D. Mathis.
1999
. Initiation of autoimmune diabetes by developmentally regulated presentation of islet cell antigens in the pancreatic lymph nodes.
J. Exp. Med.
189
:
331
.-339.
29
Turley, S., L. Poirot, M. Hattori, C. Benoist, D. Mathis.
2003
. Physiological β cell death triggers priming of self-reactive T cells by dendritic cells in a type-1 diabetes model.
J. Exp. Med.
198
:
1527
.-1537.
30
Herman, A. E., G. J. Freeman, D. Mathis, C. Benoist.
2004
. CD4+CD25+ T regulatory cells dependent on ICOS promote regulation of effector cells in the prediabetic lesion.
J. Exp. Med.
199
:
1479
.-1489.
31
Peterson, D. A., R. J. DiPaolo, O. Kanagawa, E. R. Unanue.
1999
. Quantitative analysis of the T cell repertoire that escapes negative selection.
Immunity
11
:
453
.-462.
32
Schwenk, F., U. Baron, K. Rajewsky.
1995
. A cre-transgenic mouse strain for the ubiquitous deletion of loxP-flanked gene segments including deletion in germ cells.
Nucleic Acids Res.
23
:
5080
.-5081.
33
Lacy, P. E., M. Kostianovsky.
1967
. Method for the isolation of intact islets of Langerhans from the rat pancreas.
Diabetes
16
:
35
.-39.
34
Tello, D., F. A. Goldbaum, R. A. Mariuzza, X. Ysern, F. P. Schwarz, R. J. Poljak.
1993
. Three-dimensional structure and thermodynamics of antigen binding by anti-lysozyme antibodies.
Biochem. Soc. Trans.
21
:
943
.-946.
35
Byersdorfer, C. A., R. J. Dipaolo, S. J. Petzold, E. R. Unanue.
2004
. Following immunization antigen becomes concentrated in a limited number of APCs including B cells.
J. Immunol.
173
:
6627
.-6634.
36
Anderson, M. S., E. S. Venanzi, L. Klein, Z. Chen, S. P. Berzins, S. J. Turley, H. von Boehmer, R. Bronson, A. Dierich, C. Benoist, D. Mathis.
2002
. Projection of an immunological self shadow within the thymus by the aire protein.
Science
298
:
1395
.-1401.
37
Derbinski, J., A. Schulte, B. Kyewski, L. Klein.
2001
. Promiscuous gene expression in medullary thymic epithelial cells mirrors the peripheral self.
Nat. Immunol.
2
:
1032
.-1039.
38
Hanahan, D..
1998
. Peripheral-antigen-expressing cells in thymic medulla: factors in self-tolerance and autoimmunity.
Curr. Opin. Immunol.
10
:
656
.-662.
39
Liston, A., D. H. Gray, S. Lesage, A. L. Fletcher, J. Wilson, K. E. Webster, H. S. Scott, R. L. Boyd, L. Peltonen, C. C. Goodnow.
2004
. Gene dosage–limiting role of Aire in thymic expression, clonal deletion, and organ-specific autoimmunity.
J. Exp. Med.
200
:
1015
.-1026.
40
Lohr, J., B. Knoechel, S. Jiang, A. H. Sharpe, A. K. Abbas.
2003
. The inhibitory function of B7 costimulators in T cell responses to foreign and self-antigens.
Nat. Immunol.
4
:
664
.-669.
41
Kanagawa, O., S. M. Martin, B. A. Vaupel, E. Carrasco-Marin, E. R. Unanue.
1998
. Autoreactivity of T cells from nonobese diabetic mice: an I-Ag7-dependent reaction.
Proc. Natl. Acad. Sci. USA
95
:
1721
.-1724.
42
Gagnerault, M. C., J. J. Luan, C. Lotton, F. Lepault.
2002
. Pancreatic lymph nodes are required for priming of β cell reactive T cells in NOD mice.
J. Exp. Med.
196
:
369
.-377.
43
Levisetti, M. G., A. Suri, K. Frederick, E. R. Unanue.
2004
. Absence of lymph nodes in NOD mice treated with lymphotoxin-β receptor immunoglobulin protects from diabetes.
Diabetes
53
:
3115
.-3119.
44
Shimizu, J., E. Carrasco-Marin, O. Kanagawa, E. R. Unanue.
1995
. Relationship between β cell injury and antigen presentation in NOD mice.
J. Immunol.
155
:
4095
.-4099.