Early Treatment of NOD Mice With B7-H4 Reduces the Incidence of Autoimmune Diabetes

NOD mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and housed in the Jack Bell Research Centre at the University of British Columbia. All mice were cared for according to the guidelines of the Canadian Council on Animal Care.

CD3⁺ T cells that had been purified using magnetic negative selection (StemCell Inc., Vancouver, BC, Canada) were cultured in plates that had been coated with various concentrations of anti-CD3 mAb. B7-H4.Ig or mouse Ig (10 μg/mL or indicated concentration) was cultured for 2–4 h before adding T cells.

Groups of female NOD mice were treated with B7-H4.Ig (group 1, provided by Amplimmune, Inc.) or mouse Ig (group 2) starting at the age of 3 weeks. Animals received treatment or control Ig at a dosage of 7.5 mg/kg body weight every 3 days for 12 weeks. Animals were considered diabetic after two consecutive readings with blood glucose levels of >250 mg/dL. Onset of diabetes was dated from the first sequential diabetes measurement.

The pancreata from various treatment groups were snap frozen. The cryostat sections were stained with hematoxylin and eosin. At least 100 islets were counted and blindly scored. Double staining of insulin and CD45 was developed using Texas red or FITC-conjugated secondary antibody. CD4, CD8 (BD Bioscience, Mississauga, ON, Canada), and Foxp3 (eBioscience, San Diego, CA) were stained according to the manufacturer's instructions.

Purified CD3⁺ T cells from different treatment groups were cultured with irradiated (3,000 rads) splenocytes from NOD mice and GAD p35 (20 μg/mL, amino acids 524–543) for 4 days for [³H] incorporation assay or cytokine measurement. In some experiments, CD25⁺ cells were depleted from purified CD3⁺ T cells by microbeads (Miltenyi Biotec Inc., Auburn, CA).

Single-cell suspensions were stained with fluorochrome-conjugated mAbs CD4, CD25, Foxp3, CD69, and CD44 (eBioscience). Intracellular staining of cytokine γ-interferon (IFN-γ) was performed after 4–5 h phorbol myristic acid plus ionomycin stimulation.

Islets were isolated with collagenase (Sigma, Oakville, ON, Canada) digestion of pancreata, as described previously (30). The single-cell suspension was dissociated by a nonenzymatic digestion solution (Invitrogen, Burlington, ON, Canada). These islet-infiltrating cells (IICs) were then subjected to fluorescence-activated cell sorter (FACS) analysis and real-time PCR. Real-time PCR was performed as previously described (30). The primer pairs are listed in Table 1.

The significance of the Kaplan-Meier survival curve in Fig. 2 was determined by a log-rank test. The significance of T-cell proliferation, expression of T-cell subsets, and cytokine production in Fig. 2, 3, and 4–6 was calculated by two-tailed Student t test or ANOVA. Values are expressed as means ± SEM, and differences are considered significant when P < 0.05.

B7-H4.Ig inhibits proliferation of T cells from non–diabetes-prone strains of mice such as C57BL/6 and BALB/c (26–28). To assess whether or not B7-H4 could affect the autoreactive T-cell response, we detected the effects of B7-H4.Ig on activated CD3⁺ T cells from NOD mice. Figure 1A shows that B7-H4.Ig significantly inhibited T-cell proliferation in a dose-dependent manner. T-cell proliferation was reduced by 62.2% in the presence of 10 μg/mL B7-H4.Ig (Fig. 1B). Interleukin-2 (IL-2), a hallmark of T-cell proliferation, was reduced by 60.0% (P = 0.002) (Fig. 1C) in the presence of 10 μg/mL B7-H4.Ig. The expression of early T-cell activation marker CD69 in CD4⁺ and CD8⁺ subsets was also reduced (Fig. 1D, E, F, and G). In addition, we used NRP-V7 tetramer to detect the effects of B7-H4 on the autoreactive CD8⁺ population from NOD mice. B7-H4 significantly inhibited NRP-V7 CD8⁺ cells from the NOD mice (data not shown). Collectively, these data demonstrate that B7-H4 inhibits activation of T cells from diabetes-prone NOD mice in vitro.

In order to test the effects of B7-H4 on autoimmune responses in vivo, we treated NOD mice at 3 weeks of age, presumably before or just at the onset of insulitis. Consistent with previous reports, 66.7% of the animals treated with mouse Ig (group 2, n = 24) developed hyperglycemia at between 10 and 33 weeks of age, similar to untreated animal colonies in our facility (in group 3, n = 24, the incidence of diabetes is 70.8%) (Fig. 2A). In contrast, only 28.6% of mice treated with B7-H4.Ig (group 1, n = 21) developed diabetes (Fig. 2A), significantly lower than the control groups (group 1 vs. 2, P = 0.0081; group 1 vs. 3, P = 0.0035).

To assess the kinetics of the development of diabetes in B7-H4.Ig–treated and control groups, the distribution of the onset of diabetes was compared among three groups. Diabetes developed at between 10 and 33 weeks of age in the mouse Ig–treated group, similar to the untreated group in which hyperglycemia developed at between 11 and 30 weeks of age (Fig. 2B). The onset of diabetes peaked at between 12 and 24 weeks of age in the two control groups (groups 2 and 3). In contrast, in the B7-H4–treated group, no mice developed diabetes until 13 weeks of age, 2–3 weeks later than the other two control groups, demonstrating a lower and later pattern of development of hyperglycemia with B7-H4 treatment (Fig. 2B).

Hyperglycemia in NOD mice results from destruction of insulin-producing β-cells by leukocytes that infiltrate into the islets. The reduction of diabetes by B7-H4 treatment may account for the decline of infiltrates into the pancreas. Therefore, the development of insulitis and β-cell function were examined in B7-H4– (group 1) and mouse Ig–treated (group 2) animals (Fig. 3A). Insulitis was detected in groups 1 and 2 at 4 weeks of age. However, no destructive insulitis was observed in either group (Fig. 3B). At 8 weeks of age, total infiltrated islets were accumulated to a similar degree in two groups. By 12 weeks of age, destructive insulitis was detected similarly in two groups (24 and 27% for groups 1 and 2, respectively). Notably, by 16 weeks of age, the invading intrainsulitis increased sharply to 54% in group 2, whereas it remained at 27%, a similar percentage (24%) to the 12-week results in group 1. It thus appears that severe destructive insulitis was arrested at 12 weeks of age in B7-H4–treated NOD mice.

At a late stage (16 weeks of age), we observed a distinct pattern of insulitis in two groups. Inflammation and loss of islets in the B7-H4.Ig treatment group were significantly lower compared with group 2 (inflammation, P = 0.0003; loss of islets, P = 0.0006) (Fig. 3C). There was no difference in total pancreatic insulin content and insulin secretion during an intraperitoneal glucose tolerance test (IPGTT) at 12 weeks (Fig. 3D and E). By contrast, the amount of insulin contained in the pancreata of the B7-H4–treated group at 16 weeks was significantly higher than that in controls (P = 0.003) (Fig. 3D), demonstrating a greater β-cell mass after B7-H4 treatment. Furthermore, mice in the control group demonstrated deterioration in glucose tolerance and an impaired insulin secretion during IPGTT (236 ± 56 vs. 483 ± 84 ng) at 16 weeks compared with the B7-H4–treated group (Fig. 3F). Thus, B7-H4 treatment resulted in a reduced amount of aggressive islet destruction and improved β-cell function and mass at 16 weeks. These data indicate that protection against diabetes in B7-H4–treated NOD mice was not related to blockade of insulitis. Instead, it might be associated with modulation of infiltrates at a later stage.

The distinct patterns of insulitis in NOD mice between 12 and 16 weeks of age in B7-H4 (group 1) and control (group 2) mice prompted us to investigate the phenotype of IICs. Immunohistochemistry staining showed a similar number of CD4⁺ IICs in the two groups (74 ± 20 vs. 75 ± 18 per field; P = 0.92; n = 5) (Fig. 4A). In contrast, an increased number of Foxp3⁺ (14 ± 4 vs. 22 ± 6 per field; P = 0.04; n = 5) (Fig. 4A) were stained in group 1 compared with group 2, suggesting a B7-H4–associated regulatory T-cell (Treg) phenotype at the site of inflammation. Consistent with this immunohistochemistry result on Foxp3 expression, an increased percentage of Foxp3 expression among the CD4⁺ population was found using FACS analysis (P = 0.008; n = 5) (Fig. 4B and C). A decreased number of CD8⁺ cells were stained in the pancreas (22 ± 3 vs. 13 ± 5 per field; P = 0.05; n = 5) (Fig. 4A). However, no significant alteration was found using FACS (data not shown), reflecting a fluctuation expression of CD8⁺ T-cell subsets among infiltrates.

We next examined the activation status of CD4 and CD8 subsets by measuring the early T-cell activation marker CD69. We found that B7-H4 treatment reduced CD69 expression in both CD4⁺ and CD8⁺ subsets (P = 0.04 and P = 0.02, respectively; n = 5) (Fig. 4D and E), indicating that B7-H4 inhibited activation of both the CD4⁺ and CD8⁺ population.

We next investigated the effect of B7-H4 on the function of effector cells in the pancreas. Figure 4F showed relative mRNA expression levels in IICs. A lower-level expression of IL-2, IFN-γ, granzyme B, and IL-17, and a higher-level expression of Foxp3 and IL-10 was detected in group 1 compared with group 2, suggesting that B7-H4 efficiently inhibited proinflammatory cytokines by promoting transcription of Tregs and IL-10. To test this regulation at protein levels, we performed intracellular staining of IFN-γ, which showed that the production of IFN-γ is reduced in both CD4⁺ and CD8⁺ subsets (P = 0.001 and P = 0.0005, respectively; n = 5) (Fig. 4G and H), suggesting an inhibition of cytotoxic CD8 function and a shift from T helper 1 (Th1) toward Th2 upon local B7-H4 treatment. This finding is consistent with previous evidence of a preferential inhibition of the Th1 response in vivo by B7-H4 (31).

Collectively, B7-H4 treatment resulted in a decreased Th1 response and reduced proinflammatory cytokine expression mediated by local upregulation of Tregs. This is likely to have a profound inhibitory effect on the progression from insulitis to overt diabetes.

To assess whether or not the reduction of the development of diabetes by B7-H4 occurs through the deletion of T cells or the inhibition of activated autoreactive T cells, we monitored the kinetics of CD4⁺ and CD8⁺ counts, and T-cell activation markers CD69, CD25, and CD44 at 4, 8, 12, and 16 weeks of age. No significant differences in the percentage of CD4⁺, CD8⁺, total number of T-cell subsets, CD4⁺CD69⁺, CD8⁺CD69⁺, CD4⁺CD25⁺, CD8⁺CD25⁺, CD4⁺CD44⁺, or CD8⁺CD44⁺ (data not shown) were found at any time point between the two groups, suggesting that B7-H4 neither deleted CD4⁺ and CD8⁺ T cells nor inhibited activation status in vivo in the periphery. These contradictory data may reflect the complicated microenvironment in vivo, such as the interaction between B7-H4 and other cosignaling pathways, and/or the alteration of T-cell subsets such as Th1 versus Th2 or Treg versus effector T cell (Teff).

The development of diabetes in NOD mice is the result of an imbalance of regulatory/suppressive and pathogenic/activated T cells. We showed previously that local expression of B7-H4 by a recombinant adenovirus prolongs islet allograft survival, and that this survival was associated with upregulation of Foxp3⁺ cells in the allograft (30). However, the relationship between Tregs and B7-H4 in the autoimmune response has not been elucidated. Therefore, we compared the expression of Tregs (CD4⁺CD25⁺Foxp3⁺), Teffs (CD4⁺CD25⁻Foxp3⁻), and Foxp⁺ in the CD4⁺ population (Foxp3⁺CD4⁺) of pancreatic lymph nodes (PLNs) and spleens of NOD mice in groups 1 and 2. There was no significant difference in Tregs or Teffs (data not shown). However, there was a significant increase in the percentage of Foxp3⁺CD4⁺ cells in the PLNs of 4-week-old mice treated with B7-H4.Ig compared with group 2 (P = 0.03; n = 12) (Fig. 5A and B), but not at other time points, indicating a transient regulation of suppressive Tregs in the local draining lymph nodes.

In order to assess whether or not this transient increase in the percentage of Foxp3⁺CD4⁺ cells in the PLNs is necessary for suppression of autoimmune diabetes, mixed leukocytes from diabetic NOD and 4-week-old mice treated either with B7-H4.Ig or control were adoptively transferred into NOD SCID mice. Control recipients developed hyperglycemia at an average of 7–8 weeks, whereas B7-H4–treated mice survived longer (>11 weeks; P = 0.01, by log-rank test) (Fig. 5C). This suggests that PLNs of 4-week-old NOD SCID mice treated with B7-H4.Ig displayed a greater suppressive phenotype for the development of diabetes. Although PLNs of 4-week-old mice treated with both B7-H4.Ig and control inhibited the development of hyperglycemia induced by diabetic NOD splenocytes, they showed different degrees of suppression (Fig. 5D and E). The above finding of a transient increase in the percentage of Foxp3⁺CD4⁺ cells in the PLNs may account for these differential phenotypes.

The above data, on one hand, demonstrated that inhibition of diabetes by B7-H4.Ig was not caused by a quantitative difference in the T-cell counts or activation status in the periphery. On the other hand, a transient upregulation of Foxp3⁺CD4⁺ cells in the PLNs suggested a possible alteration of T-cell function. To assess the effects of B7-H4 on the autoreactive T-cell response, we tested the proliferation of T cells in the presence of islet autoantigen GAD (Fig. 6A). T-cell proliferation was reduced by 35.8% (P = 0.02; n = 6), indicating that the ability of T cells to prime autoantigen was impaired with B7-H4 treatment.

The reduced ability of T cells to induce autoreactive T-cell proliferation in group 1 might result from a Th1/Th2 shift. We therefore examined the production of cytokines in response to autoantigens (Fig. 6B). The production of IFN-γ was reduced by 51.1% (P = 0.002; n = 6), but there was no change for IL-10 production (331 ± 96.17 vs. 292.33 ± 86.42 pg/mL; P = 0.49), indicating that the reduction of the autoreactive T-cell response might be associated with a decreased production of the Th1-related cytokine IFN-γ without affecting Th2-associated IL-10.

To assess the possibility of Treg involvement in B7-H4–mediated reduction in the autoimmune response, T-cell proliferation and cytokine production were also measured after CD25⁺ cells were depleted (Fig. 6A and B). T-cell proliferation was significantly higher after CD25⁺ depletion, reaching a similar level as the control group (P = 0.46; n = 6), suggesting involvement of Tregs in B7-H4–mediated reduced autoimmune response. Moreover, T-cell proliferation reached a similar level in both groups after CD25⁺ depletion, demonstrating a direct effect of B7-H4 on GAD p35–reactive T cells. The involvement of Tregs was further confirmed in the production of IFN-γ after depletion of CD25 (Fig. 6B).

The impaired ability for T-cell proliferation and IFN-γ production in response to GAD in group 1 may be due to either a decline in the number of cells able to respond or impaired function from each cell. To test these two possibilities, intracellular staining of IFN-γ was performed at a single-cell level. A decreased percentage of IFN-γ in both CD4⁺ (P = 0.002; n = 8) (Fig. 6C and E) and CD8⁺ (P = 0.001; n = 8) (Fig. 6D and E) was detected, suggesting a decline in the number of cells to respond to GAD.

Regulation of autoreactive T-cell responses can be achieved by several mechanisms, including costimulation blockade (32,33). The concept of controlling autoreactive T-cell responses by targeting signal 2 has been investigated in several experimental models and human studies (14,15,34). B7-H4 is a new B7 homolog molecule that interacts with an unknown receptor on activated T cells to inhibit mitogenic T-cell activation of non–diabetes-prone mouse strains. In this study, we showed that it also inhibited proliferation of T cells from the diabetes-prone NOD mouse strain. Autoreactive T-cell proliferation is thought to determine the fate of autoimmune diabetes. Blockade of endogenous B7-H4 by anti–B7-H4 mAb exacerbates EAE, an animal model of the autoimmune disease multiple sclerosis (27). Unlike EAE, which is induced by immunization with an antigen, autoimmune diabetes is a spontaneous disease caused by T cell–driven autoimmune destruction of insulin-producing β-cells. Therefore, the effect of B7-H4 in these two different autoimmune diseases and the mechanisms involved may not be the same. In this study, we found that B7-H4.Ig treatment of prediabetic mice reduced the incidence of diabetes to 28.6% compared with the control Ig group incidence of 66.7%. Our study is the first to report that B7-H4 indeed suppresses autoimmune diabetes in NOD mice.

The development of diabetes is the consequence of progressive loss of insulin-producing β-cells. Leukocytes are recruited to the pancreas and reside on the islet cells. After a long period of time, infiltrates replace the β-cells, resulting in hyperglycemia. There are two major checkpoints during the development of diabetes in NOD mice. Checkpoint 1 features the recruitment of infiltrates but free of diabetes. Leukocyte infiltrates can be stained surrounding the islets at as early as 3–4 weeks of age and accumulated further by 8–10 weeks of age (35). The second checkpoint begins at 10–12 weeks of age. At this time, nondestructive peri-insulitis is transformed into destructive intrainsulitis. β-cells are killed by dense infiltrates, which results in dysfunction of secretion of insulin from the islets. Studies into the mechanisms of the reduction of diabetes reveal that B7-H4 does not affect checkpoint 1. We observed a similar degree of insulitis in B7-H4.Ig– and control Ig–treated groups by 12 weeks of age (Fig. 3). Similar to this result, massive infiltrates are detected in allografts after islet transplantation (30). The presence of infiltrates in two different models suggests inefficient inhibition of T-cell priming by B7-H4 engagement at an early stage.

At a later stage, there is a distinct pattern of infiltration in the two groups. In controls, the severity of insulitis progresses aggressively, whereas it remains at a similar level in terms of severity of inflammation and β-cell function in the B7-H4–treated group at between 12 and 16 weeks of age, suggesting potent regulation of diabetes development by B7-H4. In NOD mice, progression from checkpoint 1 to 2 is not random or immediate. Instead, experimental data show that the transition of insulitis to overt diabetes is highly regulated over a long period of time (36–39). The autoimmune T-cell response is triggered by β-cell–specific autoantigens and may start within islets, where autoantigens originate and are presented by either β-cells or by antigen-presenting cells. However, naïve T cells cannot access peripheral nonlymphoid tissues (40). Instead, they only circulate through peripheral lymphoid organs. Early treatment of NOD females with anti-CD40 L mAb prevents insulitis and diabetes. However, it does not prevent the appearance of islet-reactive T cells, indicating that priming does not occur within the islets (41). In fact, experimental data support the idea of β-cell–specific T cells being primed in the PLNs. Several lines of evidence support this notion. First, the activation status of T cells in the PLNs is correlated with islet infiltration (42). Secondly, proliferating cells appeared in the PLNs before the onset of insulitis (42). This notion is confirmed by direct evidence of the removal of the PLNs (43). B7-H4 transiently upregulates Foxp3⁺ Tregs in the PLNs after 1 week of treatment, suggesting that the alteration of aggressiveness of insulitis in the pancreas at the late stage may be correlated with modulation of Tregs in the PLNs at a very early time point.

Autoimmune diabetes is a chronic disease. Progression of insulitis to overt diabetes is a long process. The number and/or function of Tregs in NOD mice decrease over time and are correlated with disease onset, suggesting that Tregs may be defective in NOD mice (36,44). Depletion of Tregs or loss of regulatory pathways such as CTLA-4, TGF-β, and Foxp3 promotes a more aggressive disease (19,45–47). Conversely, treatment with Tregs can delay and/or cure mice of a variety of immunological diseases (16,48,49). The site of Treg action can be peripheral or local. In the current study, our data show that B7-H4–mediated protection is associated with early upregulation of Foxp3 in the PLNs and late modulation of Tregs at the site of inflammation. At an early stage, it may correlate with a reduced autoimmune response and IFN-γ production to autoantigen GAD stimulation. At a late stage, Tregs preferentially function at the site of inflammation. Our data show that the local effect of Treg in the islets appears to limit the Th1 response, proliferation and/or cytokine production by the pathogenic T cells. Tregs in the pancreas are likely to mediate this modulation and therefore control the progression from insulitis to overt diabetes in prediabetic mice. The source of increased Tregs in the pancreas could be the result of improvement of Treg survival, or conversion of Teff into Treg, or increased recruitment of Tregs in the pancreas (37).

In the current study, we show novel data that B7-H4 inhibits T cell–mediated, spontaneously developing autoimmune diabetes in NOD mice. This result provides an additional piece of evidence to support the function of B7-H4 as a negative regulator of T-cell immunity in vivo. It has been reported that overexpression of B7-H4 prolongs graft survival and that recipients treated with B7-H4–transfected NIT-1 (an insulinoma cell line) were more resistant to STZ-induced diabetes. Survival was associated with augmented Tregs, and the production of IFN-γ and IL-4 by splenocytes was decreased and increased, respectively (50). It will be interesting to investigate the effect of B7-H4 on the modulation of antigen-presenting cells and the relationship between Treg and Th17. In conclusion, the ability of B7-H4 to negatively regulate autoimmune diabetes in NOD mice provides proof of principle for further investigation, such as whether or not early treatment with B7-H4 is necessary for prevention of disease, to what degree B7-H4 can reverse new-onset diabetes, and what the basic underlying mechanism of B7-H4 action on autoimmunity is.

This project was funded by the Canadian Institutes of Health Research (CIHR; MOP-79414). X.W. is a recipient of fellowships from the University of British Columbia Graduate Program as well as CIHR and Michael Smith Foundation for Health Research scholarship training awards in transplantation. C.B.V. is a senior scholar of the Michael Smith Foundation for Health Research.

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

X.W. researched data, contributed to discussion, and wrote and edited the manuscript. J.H. researched data and contributed to discussion. D.L.M. and A.M. contributed to discussion and edited the manuscript. Z.A. contributed to discussion. N.A. researched data. S.L., L.L., and L.C. provided new reagents and contributed to discussion. D.O. and C.B.V. contributed to discussion. G.L.W. contributed to discussion and edited the manuscript.

The authors thank Drs. Megan Leving, Joel Montance, and Erica Lee (University of British Columbia) for assistance with FACS acquisition/analysis and IIC assays.