Prevention of “Humanized” Diabetogenic CD8 T-Cell Responses in HLA-Transgenic NOD Mice by a Multipeptide Coupled-Cell Approach

Previously described NOD.β2m^null.HHD mice (17) are maintained by sibling matings at The Jackson Laboratory. MHC class I–deficient NOD.β2m^null mice have also been previously described (4). Some experiments used an N10 backcross generation NOD stock congenically expressing the CD45.2 rather than the CD45.1 variant leukocyte marker. NOD.β2m^null.HHD mice in which antigen-presenting cells (APCs) specifically express an MHC class II promoter (H2-Eα^k)–driven mouse proinsulin 2 transgene (NOD.β2m^null.HHD-PI) were generated by crossing NOD.β2m^null.HHD with previously described NOD.PI mice (27). The proinsulin transgene is maintained in a heterozygous state. The institutional Animal Care and Use Committee at The Jackson Laboratory approved all animal experiments.

Synthetic peptides Ins1/2 A_2–10 (IVDQCCTSI), Ins1 B_5–15 (HLCGPHLVEA), IGRP_228–236 (FGIDLLWSV), IGRP_265–273 (VLFGLGFAI), and Flu16 (Flu MP_58–66) (GILGFVFTL) were purchased from Mimotopes PTY, Melbourne, Australia. Monoclonal antibodies specific for CD11c (clone N418), CD8 (clone 53–6.72), CD45.1 (clone A201.7), and CD45.2 (clone 1042.1) were purchased from BD Biosciences, San Jose, CA and eBiosciences, San Diego, CA.

ECDI peptide-coupled splenocyte (peptide-SPs) treatment was carried out as previously described (28). Briefly, spleens were removed from syngeneic female mice, collagenase D treated (Roche Diagnostics, Manheim, Germany), and the erythrocytes lysed. The splenocytes were incubated with ECDI (150 mg/3.2 × 10⁸ cells [Calbiochem, La Jolla, CA]), and as indicated, a mixture of INS and/or IGRP peptide(s) (1 mg/mL each) on ice for 1 h, hand-shaking every 10 min. The peptide-SPs were washed, centrifuged, filtered to remove cell clumps, and resuspended in PBS. NOD.β2m^null.HHD female mice (4–6 weeks old) received 50 × 10⁶ peptide-SPs in 200 μL PBS by intravenous injection. Controls consisted of NOD.β2m^null.HHD female mice treated with splenocytes bearing the ECDI-coupled HLA-A2.1 binding but diabetes-irrelevant Flu16 peptide. Multiple treatments at 5-week intervals were given as indicated. The mice were monitored for diabetes development.

Diabetes onset was defined by a urinary glucose value of ≥3 as assessed with Ames Diastix (Bayer, Diagnostics Division, Elkhart, NJ). Mice were considered diabetic after two consecutive positive measurements. Mice remaining free of overt diabetes through 26 weeks posttreatment were killed and their pancreata fixed in Bouin’s solution and sectioned at three nonoverlapping levels. Granulated β-cells were stained with aldehyde fuchsin and a hematoxylin and eosin counterstain. Islets (at least 20 per mouse) were individually scored as follows: 0, no lesions; 1, peri-insular leukocytic aggregates, usually periductal infiltrates; 2, <25% islet destruction; 3, >25% islet destruction; and 4, complete islet destruction. An insulitis score for each mouse was obtained by dividing the total score for each pancreas by the number of islets examined. Overtly diabetic mice were assigned an insulitis score of 4.

Interferon (IFN)-γ enzyme-linked immunospot (ELISPOT) assays were performed as previously described (17). Spots were counted using an automated ELISPOT reader system (Immunospot; CTL Analyzers, Cleveland, OH). An interluekin (IL)-10 ELISPOT assay was also used as indicated. Numbers of IFN-γ or IL-10 spots were normalized to the indicated numbers of CD8 T-cells. NOD.β2m^null.HHD mice treated every 5 weeks with peptide-SPs were primed in a rear footpad 2 days after the third treatment with a cocktail containing 20 μg each of the HLA-A2.1–restricted Ins1/2 A_2–10, Ins1 B_5–15, IGRP_228–236, and IGRP_265–273 peptides emulsified in 50 μL of complete Freund’s adjuvant. Ten days after priming, the mice were killed, the popliteal lymph nodes were isolated and dispersed by collagenase D treatment, and CD8 T-cell reactions to each individual peptide were determined by IFN-γ ELISPOT analyses. Only mice remaining nondiabetic within each treatment group were analyzed.

Splenic CD8 T-cells were enriched from NOD.β2m^null.HHD mice by the previously described magnetic bead–based negative selection approach (29) and incubated in vitro at a concentration of 5 × 10⁵/mL with 5 × 10⁶/mL INS- and/or IGRP-SPs. The next day, the CD8 T-cells were recovered and assessed by IFN-γ ELISPOT analyses for reactivity against HLA-A2.1–restricted INS and IGRP peptides. Irradiated (2000R) syngeneic NOD.β2m^null.HHD splenocytes were used as APCs.

Starting at 3 weeks of age, NOD.β2m^null .HHD mice were treated with repeated intraperitoneal injections of a soluble mixture containing 25 μg of each peptide (Ins A_2–10, Ins B_5–15, IGRP_228–236, and IGRP_265–273). Controls consisted of NOD.β2m^null.HHD female mice treated with the Flu16 peptide. Three injections at 2-week intervals were followed by treatments once every 3 weeks. The mice were monitored for type 1 diabetes development.

Data were evaluated using Prism 5 software (GraphPad Software). The log-rank test was used to compare diabetes incidence curves and the nonparametric unpaired test to compare antigen-reactive CD8 T-cell numbers between different treatment groups.

CD8 T-cells were enriched from pooled spleens of 9- to 16-week-old NOD.β2m^null.HHD mice and cocultured in vitro with syngeneic splenocytes bearing an ECDI-coupled cocktail of the HLA-A2.1–restricted Ins1/2 A_2–10, Ins1 B_5–15, IGRP_228–236, and IGRP_265–273 β-cell autoantigens (INS/IGRP-SPs) and assessed the next day for responsiveness to each individual epitope by IFN-γ ELISPOT analyses. Compared with those exposed to the control Flu16 peptide (Flu-SPs), CD8 T-cells from NOD.β2m^null.HHD mice cocultured with INS/IGRP-SPs displayed significantly decreased responses to restimulation by IGRP epitopes (Fig. 1A). Levels of reactivity to the HLA-A2.1–restricted INS peptides was low among control CD8 T-cells exposed to Flu16 and not further influenced by previous coculture with INS/IGRP-SPs (Fig. 1A). Coincubation with syngeneic splenocytes bearing an ECDI cross-linked mixture of the two INS peptides (INS-SPs) did not diminish CD8 T-cell responsiveness to INS or IGRP epitopes (Fig. 1B). However, preincubation with IGRP-SPs did significantly diminish the ability of CD8 T-cells from NOD.β2m^null.HHD mice to respond to restimulation by IGRP but not INS epitopes (Fig. 1B). IL-10 ELISPOT analyses also indicated that regardless of preincubation conditions, no CD8 T-cells from NOD.β2m^null.HHD mice responding to INS or IGRP stimulation produced this immunosuppressive cytokine. On the basis of these collective data, we subsequently assessed syngeneic INS/IGRP-SPs as a possible diabetes intervention approach in NOD.β2m^null.HHD mice.

Initial analyses found that repeated injections of a soluble mixture of the four INS and IGRP peptides (25 μg each) did not inhibit diabetes development in NOD.β2m^null .HHD mice (data not shown). Thus, given the in vitro results shown in Fig. 1, we assessed whether a single intravenous injection of INS/IGRP-SPs given at 4–6 weeks of age could protect NOD.β2m^null.HHD female mice from diabetes development. Diabetes development was inhibited in mice treated with the INS/IGRP-SPs, compared with controls receiving Flu-SPs (Fig. 2). We next determined which peptide(s) were responsible for the diabetes protective effects. NOD.β2m^null.HHD mice were injected with syngeneic splenocytes separately bearing either a mixture of the two ECDI cross-linked INS (INS-SPs) or IGRP (IGRP-SPs) peptides. Because one injection of INS/IGRP-SPs initially afforded a partial diabetes protective effect, treatments with INS-SPs and IGRP-SPs were repeated at 5-week intervals. Compared with controls receiving Flu-SPs, only IGRP-SPs and not INS-SPs exerted a diabetes protective effect in NOD.β2m^null.HHD mice (Fig. 3). Treatments with syngeneic splenocytes separately bearing either the ECDI-coupled IGRP_228–236 or IGRP_265–273 peptide failed to significantly inhibit diabetes development in NOD.β2m^null.HHD mice (Fig. 3A). Thus, splenocytes must bear a combination of the ECDI-coupled IGRP_228–236 and IGRP_265–273 peptides in order to elicit robust diabetes protective effects in NOD.β2m^null.HHD mice. Furthermore, although both INS/IGRP-SPs and IGRP-SPs treatment could inhibit overt diabetes development in NOD.β2m^null.HHD mice, neither intervention significantly suppressed insulitis levels (Fig. 3B).

We assessed whether the diabetes protective effect of INS/IGRP-SPs treatment was associated with altered in vivo responsiveness of either HLA-A2.1–restricted INS or IGRP autoreactive CD8 T-cells in NOD.β2m^null.HHD mice. Two days following a third INS/IGRP-SPs or Flu-SPs treatment given at 5-week intervals, NOD.β2m^null.HHD mice were primed in the footpad with a cocktail of the four INS and IGRP peptides. Ten days postpriming, the mice were killed and CD8 T-cells within the draining popliteal lymph nodes were assessed for reactivity to each individual peptide by IFN-γ ELISPOT analyses. All mice were still nondiabetic at the time of analysis. In INS/IGRP-SPs–treated NOD.β2m^null.HHD mice, CD8 T-cell responses to both the HLA-A2.1–restricted IGRP_228–236 and IGRP_265–273 peptides but neither INS epitope were significantly decreased (P = 0.03 and 0.004) (Fig. 4). Treatment with IGRP-SPs but not with INS-SPs also showed significantly decreased CD8 T-cell responses to both the IGRP_228–236 and IGRP_265–273 peptides (P = 0.05 and 0.002) (Fig. 4). These results further indicate that in the context of the human HLA-A2.1 class I variant, autoreactive CD8 T-cells recognizing the IGRP_228–236 and/or IGRP_265–273 peptides are of significant pathogenic importance during diabetes development in NOD.β2m^null.HHD mice.

Previous studies have indicated that (pro)insulin is a key autoantigen for diabetes development in standard NOD mice (27,30–33). This was partly demonstrated by studies showing that MHC class II promoter–driven transgenic expression of proinsulin-2 in APCs of NOD mice inhibits insulitis and diabetes development (27,30). Hence, we used such a transgenic approach as an alternative means to test whether (pro)insulin may also be an autoantigen of pathogenic importance in NOD.β2m^null.HHD mice. NOD.β2m^null.HHD mice expressing the previously described (27) proinsulin-2 transgene in APCs (designated NOD.β2m^null.HHD-PI) were generated and assessed for diabetes development. As shown in Fig. 5A, compared with nontransgenic controls, the rate of type 1 diabetes development was somewhat retarded in NOD.β2m^null.HHD-PI female mice but did not quite achieve statistical significance (P = 0.08) by Kaplan-Meier analyses. However, as assessed by χ² analyses, the cumulative frequency of diabetes development by 35 weeks of age was significantly lower (P < 0.005) in NOD.β2m^null.HHD-PI mice than in nontransgenic controls. Insulitis levels were also significantly lower in NOD.β2m^null.HHD-PI mice than in nontransgenic controls (Fig. 5B). Baseline and primed levels of CD8 T-cell responses to the HLA-A2.1–restricted Ins1/2 A_2–10 or Ins1 B_5–14 epitopes were found not to differ in NOD.β2m^null.HHD-PI mice and nontransgenic controls (data not shown). These findings do not eliminate the possibility that like those targeting IGRP epitopes, HLA-A2.1–restricted insulin autoreactive CD8 T-cells are also important pathogenic contributors to diabetes development in NOD.β2m^null.HHD mice. However, these APC transgenic expression studies indicate that even if they do not represent autoantigens recognized by pathogenic HLA-A2.1–restricted CD8 T-cells, (pro)insulin derived epitopes are important targets of at least diabetogenic CD4 T-cells in NOD.β2m^null.HHD mice.

It has been previously reported that splenocytes bearing ECDI-coupled proteins or peptides do not directly induce tolerogenic responses by CD4 T-cells in an efficient manner, but rather do so indirectly following their uptake and processing by host-type APC (21). If this is also the case for inducing CD8 T-cell tolerance, we reasoned it should remain possible to inhibit diabetes development in the HLA-A2.1–expressing NOD.β2m^null.HHD stock by treatments with splenocytes from totally MHC class I–deficient NOD.β2m^null mice bearing the complete cocktail of ECDI cross-linked INS and IGRP epitopes (designated INS/IGRP-β2m^null SPs). Unlike those receiving Flu-β2m^null SPs, diabetes development was completely abrogated in INS/IGRP-β2m^null SPs-treated NOD.β2m^null.HHD mice (Fig. 6A).

Two days after a third INS/IGRP-β2m^null SPs or Flu-β2m^null SPs treatment given at 5-week intervals, NOD.β2m^null.HHD mice were primed in the footpad with a cocktail of the four INS and IGRP peptides. Ten days postpriming, CD8 T-cells within the draining popliteal lymph nodes were assessed for reactivity to each individual peptide by IFN-γ ELISPOT analyses. CD8 T-cell responses to both the HLA-A2.1–restricted IGRP_228–236 and IGRP_265–273 peptides but again not the INS epitopes were significantly decreased in INS/IGRP-β2m^null SPs-treated NOD.β2m^null.HHD mice (Fig. 6B). These collective results indicate that in order to induce CD8 T-cell tolerance and to elicit diabetes protective effects in NOD.β2m^null.HHD recipients, donor cells bearing ECDI cross-linked HLA-A2.1 restricted IGRP autoantigenic peptides do not also have to express the relevant host-type MHC class I variant.

The results described above also indicated that rather than directing inducing tolerogenic responses inhibiting diabetogenic CD8 cell activity in NOD.β2m^null.HHD mice, INS/IGRP-SPS or IGRP-SPs instead do so in an indirect manner dependent on their uptake by host-type APC. To directly test this possibility, we compared the ability of ECDI-treated or untreated donor cells that were also labeled with the carboxyfluorescein diacetate succinimidyl ester (CFSE) tracker dye to be taken up by host-type antigen-presenting dendritic cells (DCs). Donor- and recipient-type cells were also distinguished by respective expression of the CD45.2 versus CD45.1 pan-leukocyte marker variant. There was a significantly greater uptake by host-type splenic CD8+ DCs of ECDI-treated than untreated donor cells that had been intravenously infused 1 day earlier (Fig. 7A and B). ECDI fixation induces apoptotic death of treated cells (21). This likely explains why fewer ECDI-treated than untreated donor cells could be detected in the recipients, but with the apoptotic status of the former also allowing them to be more efficiently engulfed by host-type DCs. These results indicate the ability of donor cells bearing ECDI-coupled IGRP peptides to elicit diabetes protective effects likely entails their uptake by host-type DCs that then display the autoantigenic epitopes to CD8 T-cells in a tolerance-inducing manner.

Our results indicate that treatment with INS/IGRP-SPs (bearing an ECDI cross-linked mixture of the Ins1/2 A_2–10, Ins1 B_5–15, IGRP_228–236, and/or IGRP_265–273 peptides) inhibits diabetes development in “humanized” NOD.β2m^null.HHD mice. We found IGRP-SPs to be more effective than INS-SPs in inhibiting diabetes development in NOD.β2m^null.HHD mice. Furthermore, the inhibition of diabetes development in NOD.β2m^null.HHD mice by INS/IGRP-SPs treatment was associated with an attenuation of IGRP but not INS-specific autoreactive CD8 T-cell responses. Treatment with IGRP_228–236 or IGRP_265–273 single peptide-SPs did not significantly inhibit diabetes development in NOD.β2m^null.HHD mice. This indicated that CD8 T-cell tolerance must be established to both the HLA-A2.1–restricted IGRP_228–236 and IGRP_265–273 epitopes in order to elicit diabetes-protective effects in NOD.β2m^null.HHD mice.

In standard NOD mice, insulin appears to be an earlier target than IGRP of CD4 and CD8 T-cells initiating diabetes development (31–33). It is unclear why INS/IGRP-SPs or INS-SPs treatment did not alter levels of HLA-A2.1–restricted CD8 responses against the Ins1/2 A_2–10 or Ins1 B_5–14 epitopes in NOD.β2m^null.HHD mice. However, NOD.β2m^null.HHD mice transgenically expressing proinsulin-2 in APCs were largely protected from insulitis and diabetes development. These APC transgenic studies also indicated that even if diabetes development in NOD.β2m^null.HHD mice does not require HLA-A2.1–restricted CD8 responses targeting (pro)insulin epitopes, an important component of their disease susceptibility still entails CD4 T-cell responses against this pancreatic β-cell antigen.

It has been previously reported that although peptide-SPs can directly induce CD4 T-cell tolerance induction processes in an inefficient manner, they do so more efficiently through an indirect mechanism involving their uptake and processing by host-type APCs (25). Because of these alternative mechanisms, MHC compatibility between splenotype donor and host is not required in order to induce CD4 T-cell tolerance to ECDI-coupled antigens, although syngeneic donor cells are more efficient at doing so. We found INS/IGRP peptides ECDI coupled to completely MHC class I–deficient donor splenocytes strongly inhibited IGRP-specific CD8 T-cell responses and diabetes development in NOD.β2m^null.HHD mice. Hence, donor/host MHC class I compatibility is not required to efficiently induce CD8 T-cell tolerance to ECDI-coupled self-antigenic peptides. These findings indicate that IGRP-SPs inhibit diabetes development in NOD.β2m^null.HHD mice by inducing CD8 T-cell tolerance through an indirect host-type APC-dependent pathway. Indeed, other data indicate that host-type DCs more efficiently take up donor ECDI-treated than untreated leukocytes.

A review by Luo et al. (21) discusses efforts by the Immune Tolerance Network to develop a clinical trial using ECDI insulin-coupled peripheral blood lymphocytes as a possible diabetes intervention in humans. Our current results indicate that the use of “humanized” NOD.β2m^null.HHD mice and other related strains may facilitate the development of clinically translatable peptide-based therapies for diabetic patients. In particular, currently available “humanized” mouse resources make it possible to determine which autoantigenic peptides when ECDI cross-linked to autologous leukocytes are most likely to attenuate HLA-A2.1–restricted CD8 T-cell responses that recent evidence (10–16) indicates may be important for diabetes development in many human patients.

This work was supported by National Institutes of Health grants DK-46266 and DK-51090, and Cancer Center Support Grant CA34196, and grants from the Juvenile Diabetes Research Foundation (JDRF). M.N. is the recipient of a JDRF post-doctoral fellowship award.

No potential conflicts of interest relevant to this article were reported.

M.N. researched data and wrote the manuscript. A.E.G. researched data. M.M. created the NOD.β2m^null.HHD mice. T.W.H.K. created NOD.PI mice and contributed to discussion. D.L.G. contributed to discussion and reviewed and edited the manuscript. D.V.S. directed research and wrote the manuscript.