Transfusion of Apoptotic β-Cells Induces Immune Tolerance to β-Cell Antigens and Prevents Type 1 Diabetes in NOD Mice

Female NOD mice, NOD-Rag^−/− mice, and NOD.BDC2.5 mice were purchased from The Jackson Laboratory and housed in a specific pathogen–free facility of the Mouse Colony of the Department of Pathology, Immunology and Laboratory Medicine at the University of Florida. The animal studies followed the guidelines of the University of Florida Institutional Animal Care and Use Committee.

The NIT-1 cell line was a gift of Dr. David Serreze (The Jackson Laboratory) and was maintained in RPMI-1640 medium with 10% FCS. Culture media were changed every 3 days, and the cell passage was performed once a week through trypsinizing procedure.

For preparation of UVB-irradiated NIT-1 cells, NIT-1 cells were irradiated with UVB (1,200 mJ/m²) in a 3-cm Petri dish. After irradiation, the irradiated NIT-1 cells were harvested and enumerated with a hemacytometer under a microscope.

Splenocytes (1 × 10⁶) with or without CD25⁺ cell depletion from NOD mice receiving different treatments as indicated were stimulated with 12.5 μmol/l insulin B9-23 peptide (Peptides International, Louisville, KY) in 200 μl RPMI-1640–10% FCS in a 96-well plate for 5 days. [³H]Thymidine was added to each well (1 μCi/well) and incubated for additional 16 h. T-cell proliferation was determined by scintillation counting. In the experiments for preparing supernatants for cytokine assay, splenocytes (1 × 10⁶) were stimulated with 12.5 μmol/l insulin B9-23 for 5 days, and the supernatants were harvested and frozen at −80°C for later cytokine measurement.

The splenocytes from 4-week-old NOD.BDC2.5 mice treated with three weekly transfusions of UVB-irradiated NIT-1 cells (1 × 10⁵ cells/mouse) or PBS were stimulated with 12.5 μmol/l T-cell receptor (TCR)-specific peptide 1040-55 (Peptides International, Louisville, KY) for 5 days. Then, the cells were stimulated with phorbol myristic acid (50 ng/ml) and ionomycin (1 μg/ml) in the presence of GolgiStop reagent (B-D PharMingen, San Diego, CA) for 4 h. The cells were stained with CD4-PerCP first and then with anti–interferon (IFN)-γ fluorescein isothiocyanate and anti–interleukin (IL)-10 allophycocyanin (APC) using an intracellular cytokine staining kit (B-D PharMingen). The cytokine-secreting CD4⁺ T-cells were analyzed by flow cytometry.

For Foxp3 staining, the spleen cells were stained with fluorescence-conjugated anti-CD4 antibody combined with anti-CD25 antibody (B-D PharMingen). Thereafter, the cells were fixed and permeabilized using an intracellular staining kit (B-D PharMingen), and then stained with anti-Foxp3 (eBioscience) or isotype control antibody (eBioscience). The CD4⁺CD25⁺ cells and Foxp3-expressing cells were analyzed by flow cytometry.

Four-week-old NOD mice were treated with three weekly transfusions of UVB-irradiated NIT-1 cells (1 × 10⁵ cells/mouse) or PBS (three mice in each group). Thereafter, all the mice received an antigenic challenge by intraperitoneal injection of 50 μg NIT-1 lysates combined with 100 μl adjuvant Alum Imject (Pierce Biotechnology, Rockford, IL) once a week for 2 weeks. In the following week, the sera of all mice were prepared for anti–β-cell antibody assays.

NIT-1 cells were fixed and permeabilized using an intracellular cytokine staining kit (B-D PharMingen). Then, 1 × 10⁵/20 μl of the above NIT-1 cells were incubated with 20 μl sera as indicated at room temperature for 30 min. After two washes with PBS, the cells were incubated with phycoethrin-conjugated anti-mouse Ig (B-D PharMingen) for 30 min. Anti–β-cell antibodies were examined by flow cytometry. Mean fluorescence intensity (MFI) was analyzed for each serum sample. The serum samples from naive 4-week-old NOD mice were used as isotype controls.

T-cell cytokines in the supernatants prepared above were measured by Luminex. The cytokine assay kits for mouse IL-4, IL-10, and IFN-γ were purchased from Upstate USA (Charlottesville, VA). All the procedures were performed according to the instructions from the manufacturer.

Ten-week-old NOD mice were chosen for use in these experiments. NOD mice received three weekly transfusions of UVB-irradiated NIT-1 cells, nonirradiated NIT-1 cells (1 × 10⁵ cells/mouse), or PBS. Urine glucose was monitored once a week. Once urine glucose became positive, diabetes was confirmed in the next 2 consecutive days by measuring both urine and blood glucose.

NOD-Rag^−/− mice served as recipient mice. NOD-Rag^−/− mice are nonleaky severe combined immunodeficient mice and are frequently used as the recipients for adoptive transfer experiments in type 1 diabetes research. NOD-Rag^−/− mice received splenocytes (1 × 10⁷/mouse) from diabetic NOD mice along with splenocytes (1 × 10⁷/mouse) from the NOD mice having received three weekly transfusions of UVB-irradiated NIT-1 cells (1 × 10⁵/mouse) or PBS. The onset of diabetes was monitored as described above.

Four-week-old NOD mice were treated with UVB-irradiated NIT-1 cells or PBS by three weekly transfusions. The following week, all the mice received intravenous injections of 1 × 10⁷ carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled splenocytes from NOD-BDC2.5 mice, which are β-cell antigen–specific TCR-transgenic mice with their TCRs on CD4⁺ T-cells recognizing β-cell antigens. At the same time, intraperitoneal injection of streptozotocin (40 mg/kg) was given to each mouse to ensure similar β-cell antigenic exposure to the injected splenocytes in all mice. Four days later, CD4⁺ T-cell proliferation of CFSE-labeled cells from different lymphoid tissues was examined by flow cytometry after CD4 staining with anti-CD4-phycoethrin antibody.

A Student's t test was applied for analyzing the differences of cytokine production and the proliferation of splenocytes between different groups. A log-rank test was used for comparing incidences of diabetes in different groups. The difference was considered to be significant when P < 0.05.

To determine whether transfusion of apoptotic β-cells induces β-cell antigen-specific tolerance in terms of suppression of anti–β-cell antibodies, the following experiments were performed. The NOD β-cell line (NIT-1) was chosen for use in our experiments. The NIT-1 cells were induced to undergo apoptosis by UVB irradiation (see Fig. 1 of the online appendix available at http://dx.doi.org/10.2337/db06-0825). UVB-irradiated NIT-1 cells (1 × 10⁵/mouse) were transfused into recipient NOD mice (4 weeks old) once a week for 3 weeks. Thereafter, all of the mice received two weekly challenges by NIT-1 lysates. We found that the serum levels of anti–β-cell antibodies were significantly lower in UVB-irradiated NIT-1 cell–treated mice than in PBS-treated mice after β-cell antigenic challenges (Fig. 1), suggesting that the development of antibodies was suppressed by UVB-irradiated NIT-1 cell treatment. In addition, the correlation between MFI values and radioimmunoassay index values for insulin-associated autoantibodies (flow cytometry versus radioimmunoassay) suggests that serum insulin–associated autoantibodies might have contributed to the increased MFIs in the PBS-treated group (Table 1 of the online appendix).

To assess the T-cell response induced by the transfusion of UVB-irradiated NIT-1 cells, we stimulated the splenocytes from the mice described above with insulin B9-23, which was demonstrated to be a major target antigen (23,24) presented by major histocompatibility complex-II (I-A^g7) molecules and only stimulating CD4⁺ T-cells as shown in Fig. 2 of the online appendix. In some experiments, CD25⁺ cells were depleted from the splenocytes to determine whether β-cell antigen–specific regulatory T-cells were induced. We found that the proliferation of splenocytes from UVB-irradiated NIT-1 cell–treated mice was significantly lower than that in control mice (Fig. 2A) (P < 0.01). We also found that the majority of CD25⁺ cells were in the CD4⁺ cell population, and the percentages of CD4⁺CD25⁺ T-cells were similar in both groups (Fig. 2B). Of interest, depletion of CD25⁺ cells promoted the proliferation of the splenocytes from UVB-irradiated NIT-1 cell–treated mice, whereas the same cell depletion attenuated the proliferation of the splenocytes from PBS-treated mice. These results suggest that the CD25⁺ cells in the CD4⁺ T-cell population behave differently in the two groups (UVB-irradiated NIT-1 cells versus PBS). In agreement with the above findings, the results shown in Fig. 3A of the online appendix demonstrated that splenic CD4⁺CD25⁺ T-cells from UVB-irradiated NIT-1 cell–treated mice significantly inhibited insulin B9-23–stimulated CD4⁺ T-cell proliferation compared with those from PBS-treated mice. We also found that insulin B9-23–stimulated proliferation of T-cells from pancreatic draining lymph nodes (PLNs) of UVB-irradiated NIT-1 cell–treated mice was significantly suppressed compared with that in PBS-treated mice (Fig. 3B of the online appendix). It was also noted that CD4⁺CD25⁺ T-cells from the mice with different treatments expressed different levels of Foxp3 (Fig. 2B).

We found that the splenocytes from UVB-irradiated NIT-1 cell–treated mice produced levels of IL-4 and IL-10 that were significantly higher and a level of IFN-γ that was comparable (Fig. 3A) to that in PBS-treated mice. The extremely high levels of IL-10 production in UVB-irradiated NIT-1–treated mice (Fig. 3A) suggest that IL-10–producing T-regulatory type 1 (Tr1) cells (25) might be induced. In addition, we found that NIT-1 lysates (Fig. 4 of the online appendix) stimulated the cytokine-secreting patterns similar to that induced by insulin B9-23. Intracellular cytokine staining results from BDC2.5 mice confirmed the fact that UVB-irradiated NIT-1 cell treatment induces IL-10–and IFN-γ–producing Tr1 cells (Fig. 3B).

To assess the efficacy of this treatment in preventing type 1 diabetes, we chose to treat 10-week-old NOD mice in the pre-diabetes period (26). It is thought that the disease process in diabetes-prone NOD mice at this stage is difficult to halt (27) unless the intervention is highly effective. In most studies, diabetes-prone NOD mice start to develop diabetes after 10 weeks of age (26). To our surprise, three transfusions of UVB-irradiated NIT-1 cells significantly delayed the onset of type 1 diabetes (Fig. 4). By 20 weeks of age, 75% (9 of 12) of the PBS-treated mice developed diabetes, whereas only 23% (3 of 13) of the UVB-irradiated NIT-1 cell–treated mice were diabetic (P < 0.01). By 30 weeks of age, 100% (12 of 12) mice in the PBS group versus 61.5% (8 of 13) in the UVB-irradiated NIT-1 cell–treated group were diabetic. These results demonstrated that transfusion of UVB-irradiated NIT-1 cells significantly delayed and prevented the onset of type 1 diabetes in NOD mice treated at the late stage of insulitis. In addition, the treatment with NIT-1 cells without UVB irradiation was not effective in preventing diabetes (Fig. 4), suggesting that the protective effect of UVB-irradiated NIT-1 cell treatment was not due to the insulin produced by the few alive NIT-1 cells probably included in the injected UVB-irradiated NIT-1 cells.

To further determine whether regulatory T-cells were induced by transfusion of UVB-irradiated NIT-1 cells, adoptive cotransfer experiments were performed. The results shown in Fig. 5 demonstrate that the splenocytes from the mice that received UVB-irradiated NIT-1 cell treatment protected NOD-Rag^−/− mice from developing diabetes caused by simultaneous infusion of diabetogenic NOD splenocytes.

To further determine whether UVB-irradiated NIT-1 cell treatment induces β-cell-antigen-specific immunosuppression, we treated NOD mice (4 weeks old) with three weekly transfusions of UVB-irradiated NIT-1 cells or PBS. The week after the last transfusion, we intravenously injected CFSE-labeled splenocytes from NOD.BDC2.5 mice with their TCRs on CD4⁺ T-cells recognizing islet antigenic peptides (28,29). To ensure a similar exposure to β-cell antigens for NOD.BDC2.5 splenocytes injected to different group of mice, we induced β-cell damage by intraperitoneal injection of streptozotocin (40 mg/kg) right after intravenous injection of splenocytes. The results shown in Fig. 6 demonstrate that the divisions of CFSE-labeled NOD.BDC2.5 CD4⁺ T-cells in PLNs were fewer in UVB-irradiated NIT-1 cell–treated mice than in PBS-treated mice, suggesting that NOD.BDC2.5 CD4⁺ T-cell proliferation in PLNs stimulated by endogenous β-cell antigens was suppressed in UVB-irradiated NIT-1 cell–treated mice. There was similar and little proliferation of CFSE-labeled CD4⁺ T-cells of other lymphoid tissues in both groups (Fig. 6).

Although the effect of β-cell apoptosis on the development of type 1 diabetes is still controversial (30–32), the fact that in vivo induction of β-cell apoptosis effectively prevents type 1 diabetes in NOD mice (22) suggests that application of apoptotic β-cells induced ex vivo to prevent type 1 diabetes is possible. Furthermore, from a practical standpoint, ex vivo–induced apoptotic β-cells would be more useful and feasible for future clinical applications. In the present study, we demonstrated an effective approach to prevent type 1 diabetes by transfusions of apoptotic β-cells (NIT-1 cells) induced by UVB irradiation.

It has been known that UVB irradiation induces cell apoptosis (33) that, in turn, contributes to the immunosuppression caused by UVB irradiation (33–35). Normal tissue turnover in the replacement of apoptotic cells by the immune system without causing autoimmunity suggests that steady-state apoptotic cells have an immunoregulatory function through direct or indirect pathways (11,36,37). Steinman et al. (38) have proposed a paradigm for how self-tolerance is maintained by immature dendritic cells, i.e., steady-state immature dendritic cells in the periphery pick up the apoptotic cells during normal tissue turnover and migrate to the local lymph nodes whereby they tolerize self-reactive CD4⁺ and CD8⁺ T-cells. Recent evidence shows that after interaction with apoptotic cells, dendritic cells acquire IL-10–producing capacity but lose IL-12–producing capacity (37), suggesting that apoptotic cells suppress the immune response probably through the effects of dendritic cells. In our study, we found that the dendritic cells and macrophages could pick up the injected UVB-irradiated NIT-1 cells in vivo (data not shown). Therefore, those APCs may present β-cell antigens to tolerize β-cell antigen–reactive T-cells through the mechanisms discussed above. To ensure that the transfused UVB-irradiated NIT-1 cells were apoptotic with good cell integrity, we performed the cell transfusions within 2 h after UVB irradiation.

We found that 2 h after UVB irradiation, NIT-1 cells underwent typical morphological features of apoptosis. By 4 h, >80% of the cells were apoptotic as defined by flow cytometry (Fig. 1 of the online appendix). The suppression of β-cell antibody development in UVB-irradiated NIT-1 cell–treated mice suggests that humoral immune tolerance is induced (Fig. 1 and Table 1 of the online appendix). It is likely that the suppression of B-cell function is through immunosuppressive regulatory T-cells as described recently (39,40). Indeed, the islet β-cell antigen–specific regulatory T-cells were induced by UVB-irradiated NIT-1 cell treatment (Figs. 2 and 3). In our experiments, we found that the splenocytes from UVB-irradiated NIT-1 cell–treated mice receiving two β-cell antigenic challenges had significantly lower proliferation upon in vitro stimulation of insulin B9-23 than the splenocytes from PBS-treated mice that also received two β-cell antigenic challenges (Fig. 2A). Similar results were obtained when the cells from PLNs were stimulated with insulin B9-23 (Fig 3B of the online appendix). We also found that the percentages of CD4⁺CD25⁺ T-cells were similar in both groups (Fig. 2B). However, of interest, depletion of CD25⁺ cells promoted the proliferation of the splenocytes from UVB-irradiated NIT-1 cell–treated mice, whereas the same cell depletion attenuated the proliferation of the splenocytes from PBS-treated mice. These findings suggest that UVB-irradiated NIT-1 cell treatment induces CD25⁺ regulatory T-cells and antigenic challenge does not activate effector cells, whereas in PBS-treated mice β-cell antigen-specific effector cells might be activated by antigenic challenge and express CD25 as an activation marker, therefore depletion of CD25⁺ cells attenuates proliferation. As shown in Fig. 2B, the majority of CD25⁺ cells in splenocytes were CD4⁺, which suggests that UVB-irradiated NIT-1 cell treatment induces β-cell antigen-specific CD4⁺CD25⁺ regulatory T-cells. Similar results were obtained when splenocytes were stimulated by NIT-1 lysates (Fig. 4 of the online appendix), suggesting that UVB-irradiated NIT-1 cell treatment induces immune tolerance to multiple β-cell antigens. Consistent with the above finding, CD4⁺CD25⁺ T-cells from UVB-irradiated NIT-1 cell–treated mice expressed higher levels of Foxp3 than those from PBS-treated mice (Fig. 2B). Foxp3 expression is critical for regulatory T-cells to perform their immunosuppressive function (14,41). Standard immunosuppression assays demonstrated that purified splenic CD4⁺CD25⁺ T-cells from the UVB-irradiated NIT-1 cell–treated mice but not the PBS-treated mice significantly suppressed insulin B9-23–stimulated T-cell proliferation (Fig. 3A of the online appendix), further indicating that UVB-irradiated NIT-1 cell treatment induced immunosuppressive regulatory T-cells.

Further study demonstrated that the splenocytes from UVB-irradiated NIT-1 cell–treated mice produced significantly higher levels of IL-4 and IL-10 and a comparable level of IFN-γ (Fig. 3) than those from PBS-treated mice, indicating that UVB-irradiated NIT-1 cell treatment induces Th2 and perhaps IL-10–and IFN-γ–producing Tr1 cells (25). Intracellular cytokine staining further confirmed that IL-10–and IFN-γ–producing Tr1 cells were induced by UVB-irradiated NIT-1 cell treatment, which was supported by a recent report (42) showing that induced regulatory T-cells produced IFN-γ and IL-10. Accordingly, the IFN-γ produced by splenocytes from UVB-irradiated NIT-1 cell–and PBS-treated mice was not from the same type of T-cells: the former was mainly from Tr1 cells and the latter was predominantly from Th1 cells (Fig. 3). Recent evidence shows that IFN-γ plays an important role for regulatory T-cells to exert their functions (42–44). Therefore, the production of IFN-γ in UVB-irradiated NIT-1 cell–treated mice may play an important role in facilitating the formation of immune tolerance through regulating regulatory T-cell function.

The encouraging results discussed in this article prompted us to determine whether UVB-irradiated NIT-1 cell treatment was able to prevent type 1 diabetes in NOD mice. To assess its efficacy, we chose to use 10-week-old NOD mice with a well-established autoimmune process (26,27). Our results demonstrated that transfusion of UVB-irradiated NIT-1 cells significantly delayed and prevented the onset of type 1 diabetes in NOD mice treated at the late stage of insulitis. The mechanisms underlying this protective effect are not yet clear. The induced β-cell antigen–specific regulatory T-cells could have been involved in this protection, as suggested by the results shown in Figs. 1, 2, and 3. Recent evidence shows that apoptotic cells can suppress IL-12 production by APCs, such as macrophages and dendritic cells (37), which may also play a role in tolerance induction. It has been demonstrated that phosphatidylserine on the cell surface of apoptotic cells or soluble phosphatidylserine can suppress immune responses (10,11,37,45). The immunosuppression by phosphatidylserine on apoptotic UVB-irradiated NIT-1 cells could have contributed to the protective effect of UVB-irradiated NIT-1 cell treatment through an antigen-nonspecific fashion. Recently, Tisch's group reported that apoptotic cells induced Mer tyrosine kinase–dependent blockade of nuclear factor-κB activation and suppressed proinflammatory cytokine production by dendritic cells (46); therefore, the dendritic cells modulated by apoptotic NIT-1 cells might participate in the tolerance induction. It should be pointed out that in mechanism studies we used 4-week-old mice, whereas in the type 1 diabetes prevention studies, we chose to use 10-week-old NOD mice with advanced insulitis to evaluate how efficacious this treatment was. The mechanisms underlying the tolerance induced in NOD mice with different stages of insulitis (4 vs. 10 weeks) by UVB-irradiated NIT-1 cell treatment might not be the same. However, three weekly transfusions of UVB-irradiated NIT-1 cells starting from 10 weeks of age significantly prevented diabetes in NOD mice, suggesting that β-cell antigen–specific immune tolerance was induced. One or more mechanisms discussed above regarding tolerance induction by apoptotic cells could have been involved. It was also noted that live NIT-1 cells without UVB irradiation were not effective in preventing type 1 diabetes (Fig. 4), which suggests that NIT-1 cells must undergo apoptosis to induce immune tolerance to β-cell antigens and thereby prevent type 1 diabetes. This finding also suggests that NIT-1 cells were not the “concealed” insulin producers that were responsible for the effectiveness of type 1 diabetes prevention by UVB-irradiated NIT-1 cell treatment. Moreover, we found that the insulin level in the culture with 1 × 10⁵ UVB-irradiated NIT-1 cells in 1 ml medium for 48 h was almost undetectable (data not shown), further suggesting that the little insulin released from UVB-irradiated NIT-1 cells played no role in the euglycemia in the mice protected by UVB-irradiated NIT-1 cell treatment.

From the diabetes development curve (Fig. 4), it is noted that three weekly transfusions of UVB-irradiated NIT-1 cells markedly slowed the disease process. However, some of the mice gradually developed diabetes at late time points. This result suggests that the strategy of using UVB-irradiated β-cells to prevent type 1 diabetes needs to be modulated. More injections or modified injection timing might be needed.

The results of adoptive cotransfer shown in Fig. 5 confirm that UVB-irradiated NIT-1 cell treatment induces disease-protective regulatory T-cells. Furthermore, we found that the proliferation of injected β-cell antigen-specific CD4⁺ T-cells of NOD.BDC2.5 mice was markedly suppressed in PLNs of UVB-irradiated NIT-1 cell–treated mice (Fig. 6), indicating that UVB-irradiated NIT-1 cell treatment induces immune tolerance to multiple β-cell antigens.

In conclusion, the transfusion of UVB-irradiated apoptotic β-cells effectively prevents type 1 diabetes in NOD mice through induction of β-cell antigen–specific immune tolerance. This promising approach has great potential for management of human type 1 diabetes because the existing evidence has initially shown that type 1 diabetic patients benefit from photopheresis therapy (47).

This work was supported by an American Diabetes Association Junior Faculty Award and a Juvenile Diabetes Research Foundation award to C.Q.X. and partially supported by National Institutes of Health Grant R21DK063422 to M.C.S.