Inhibition of Autoimmune Diabetes by Oral Administration of Anti-CD3 Monoclonal Antibody

AKR/J male mice (5–6 weeks old) were purchased from The Jackson Laboratory and housed in a pathogen-free animal facility at the Harvard Institutes of Medicine according to the animal protocol guidelines of the Committee on Animals of Harvard Medical School. For recovery of anti-CD3 mAb after feeding, SJL mice were used (The Jackson Laboratory).

Mice were injected intraperitoneally for 5 consecutive days with 40 mg/kg body weight of STZ (Sigma Aldrich, St. Louis, MO) in 0.01 mol/l citrate buffer (pH 4.5) to induce diabetes. Day 1 was defined as the first STZ injection. An intraperitoneal glucose tolerance test (IPGTT) was performed to diagnose diabetes using a glucose meter (Hypoguard, Minneapolis, MN). Blood glucose levels were estimated from the tail vein 30 min after an intraperitoneal injection of glucose (1.33 g/kg body weight). Mice with a glucose level of >250 mg/dl for ≥2 consecutive weeks were considered to be diabetic.

Anti-CD3 mAb, whole antibody (Ab), and (145-2C11) F(ab′)₂ were purchased from Bio Express (West Lebanon, NH). Hamster IgG whole Ab and F(ab′)₂ were purchased from Jackson ImmunoResearch Laboratories (Charlestown, MA) and were used as control Ig. Mice were fed with anti-CD3 mAb, control IgG, or PBS followed by STZ injection (40 mg/kg) daily for 5 consecutive days (days 1–5). In combination with anti-CD3 mAb feeding in some experiments, mice were injected with 500 μg neutralizing anti–TGF-β Ab (Bio Express) or mouse IgG (Jackson ImmunoResearch Laboratories) as control Ig on days 0, 2, 4, 6, and 14.

For histological examination of pancreatic islets, the pancreas was removed on day 28 and fixed in 10% buffered formalin. Specimens embedded in paraffin wax were cut by microtome, and sections were stained with hematoxylin and eosin. The slides were analyzed for presence of insulitis by an individual without knowledge of the identity of samples.

Single-cell suspensions were prepared from spleen and mesenteric lymph node (MLN). To measure proliferation, cells were plated in 96-well round-bottom plates (2.5 × 10⁵ cells/well) in cRPMI containing: RPMI 1640 medium with 10% FBS (BioWhittaker, Walkersville, MD), 100 units/ml penicillin (BioWhittaker), 100 μg/ml streptomycin (BioWhittaker), and 10 mmol/l HEPES (BioWhittaker) and stimulated with 1 μg/ml soluble anti-CD3 mAb (BD Pharmingen, San Diego, CA). After 48 h in culture at 37°C with 7% CO₂, 1 μCi [³H]thymidine was added to each well, and cells were harvested 12 h later. Proliferation was detected using a 1450 Microbeta liquid scintillation counter (PerkinElmer, Boston, MA). For cytokine production, cells were cultured at 2.5 × 10⁵ cells/well in X-VIVO 20 medium (BioWhittaker) and stimulated with 1 μg/ml anti-CD3 mAb. The culture supernatants were collected at 48 h for interleukin (IL)-2, IL-4, IL-10, and interferon (IFN)-γ and at 72 h for TGF-β. Enzyme-linked immunosorbent assays for IL-2, IL-4, IL-10, and IFN-γ were performed as reported previously (20). TGF-β was assayed using a mouse/rat/porcine TGF-β1 immunoassay kit according to the manufacturer's protocol (R&D Systems, Minneapolis, MN).

Flow cytometry analysis was performed on single-cell suspensions isolated from MLN, spleen, and pancreatic lymph node (PLN). Cells were stained in PBS with 2% BSA and fixed with 1% paraformaldehyde. Flow cytometry analysis was performed by a FACScan using Cellquest software. The antibodies used were anti-CD16/CD32 mAb (2.4G2) as the Fc blocker, fluorescein isothiocyanate-conjugated anti-CD4 mAb (L3T4), phycoerythrin (PE)-conjugated anti-CD3 mAb (145 2C11), PE-conjugated anti-CD25 mAb (7D4), antigen-presenting cell (APC)-labeled streptavidin, PE-conjugated anti-CD62L Ab (MEL14), PE-conjugated anti-CD45RB Ab (16A) (BD Pharmingen), and biotinylated anti-human LAP Ab (BAF246; XO02) (R&D Systems).

CD4⁺ T-cells from spleen cells were purified using a mouse T-cell CD4 enrichment column kit (R&D Systems) according to the manufacturer's instructions. LAP⁺CD4⁺ cells were purified by immunomagnetic cell sorting (MACS) (Miltenyi Biotec, Auburn, CA). All procedures were performed in accordance with the manufacturer's instructions. Purified CD4⁺ T-cells were stained with biotinylated anti-human LAP Ab and streptavidin-magnetic microbeads in the presence of Fc block. Then LAP⁺CD4⁺ cells were positively separated and LAP⁻CD4⁺ cells negatively separated by the MACS column. In some experiments, CD25⁺ cells were excluded from LAP⁻CD4⁺ cells or CD4⁺ T-cells followed by staining with biotinylated anti-human LAP Ab and streptavidin-magnetic microbeads. Cells stained in a combination of PE-conjugated anti-CD25 Ab and anti-PE magnetic microbeads were separated into CD25⁺ cells and CD25⁻ cells by the MACS column. To prepare APCs depleted of T-cells, spleen cells from naive mice were stained with a combination of PE-conjugated anti-CD3 mAb and anti-PE magnetic microbeads, and stained cells were negatively isolated by the MACS column. Purified cells were used for cytokine assays or suppressive function assays. For cytokine assays, 2.5 × 10⁵ LAP⁺CD4⁺ cells, LAP⁻CD25⁺CD4⁺ cells, or LAP⁻CD25⁻CD4⁺ cells were cultured with 10 μg/ml plate-bound anti-CD3 mAb in X-VIVO 20 (BioWhittaker) in 96-well round-bottom plates. The culture supernatants were collected at 48 h for IL-2, IL-4, IL-10, and IFN-γ and at 72 h for TGF-β. For measuring suppressive function in a coculture system, titrated purified cells were cultured with an equal ratio of responder cells (LAP⁻CD25⁻CD4⁺ cells) and APCs and stimulated with 1 μg/ml soluble anti-CD3 mAb in cRPMI. Cocultured cells were maintained for 60 h, and 1 μCi [³H]thymidine was added at 12 h before harvest. In some experiments, 50 μg/ml neutralizing Ab, anti–IL-10 Ab (JES052A5), or anti–TGF-β Ab (1D11), purchased from R&D systems, was added to the culture system.

One day after the last feeding, CD4⁺ T-cells or LAP⁻CD4⁺ cells were purified from spleen and MLN of mice fed with 50 μg anti-CD3 mAb or control IgG for 5 consecutive days. The purified cells (2.5 × 10⁶) were injected intravenously to naive AKR/J mice on day 0. STZ (40 mg/kg) was administrated to the recipient AKR/J mice from days 1 to 5, and an IPGTT was performed on days 7, 14, 21, and 28 using a glucose meter.

Six hours after feeding, the upper part of the small intestine from three mice fed 50 μg anti-CD3 mAb or control isotype Ab was digested using 1 ml PBS containing 10 mg collagenase D (Roche) for 30 min at 37°C. The tissue was homogenized and centrifuged for 30 min at 70,000 rpm. After centrifugation, the soluble fraction was added to a protein-G column, and the bound IgG was eluted in the presence of proteinase inhibitors (Roche).

The biological activity and the concentration of the eluted IgG were ascertained by the ability to induce T-cell proliferation of naive splenocytes compared with known concentrations of anti-CD3 mAb. The culture conditions for proliferation with anti-CD3 mAb were as follows. Single cell suspensions were prepared from spleens of naive SJL mice. Cells were plated in 96-well round-bottom plates (2.5 × 105 cells/well) in X-VIVO 20 and stimulated with either soluble known concentrations of anti-CD3 mAb (0.0025, 0.005, or 0.01 μg/ml) or with 0.5- to 5-μl eluates. After 72 h in culture 1 μCi [³H]thymidine was added to each well, and cells were harvested 12 h later. Each experiment was repeated twice. The statistic used was an unpaired Student's t test.

Statistical significance was calculated using ANOVA with Bonferroni post hoc analysis, the unpaired Student's t test assuming unequal variance, or a χ² test for in vitro assays. For in vivo assays the Mantel-Cox log-rank test or a χ² test was used.

Mice were orally administered 5, 50, or 250 μg anti-CD3 mAb and injected with STZ to induce diabetes on days 1–5. Thereafter, blood glucose levels were monitored once a week, and mice with a glucose level >250 mg/dl by IPGTT for 2 or more consecutive weeks were classified as diabetic. As shown in Fig. 1, diabetes progression was not suppressed in mice fed with 5 μg anti-CD3 mAb compared with the PBS group or control IgG group. On the other hand, mice treated with 50 or 250 μg anti-CD3 mAb had significantly reduced development of diabetes compared with mice treated with either PBS or control IgG. On day 14, diabetes was diagnosed in 67% (8 of 12) of mice treated with PBS and 83% (10 of 12) of mice treated with control IgG. However, development of diabetes in mice treated with anti-CD3 mAb was only 33% (4 of 12). Mice treated with 50 μg anti-CD3 mAb were followed for up to 42 days. At this time, there was no statistically significant difference between the treated and control groups although there were fewer diabetic animals in the treated group (8 of 12) than in the control group (10 of 12). Significant suppression of diabetes was also observed in animals fed 250 μg anti-CD3 mAb. Correspondingly, as shown in Fig. 1D, the blood glucose levels on day 14 were also significantly reduced in mice treated with 50 μg anti-CD3 mAb compared with mice treated with either PBS or control IgG. Moreover, as shown in Table 1, the mice treated with 50 μg anti-CD3 mAb had a significantly lower incidence of insulitis in the pancreas than that of mice treated with either control IgG or PBS on day 28.

It has been reported that IV anti-CD3 mAb can treat diabetes in NOD mice and that the effect is related to the immunoregulatory function of TGF-β (7). Therefore, we investigated whether the regulatory function in the periphery was enhanced in the STZ model after 50 μg oral anti-CD3 mAb treatment and STZ injection. On day 6, spleens were removed from each group, and cells were cultured with soluble anti-CD3 mAb to measure cell proliferation and cytokine production. As shown in Fig. 2, spleen cells from mice treated with either PBS or control IgG proliferated to anti-CD3 mAb in a dose-dependent manner. On the other hand, proliferation of spleen cells from mice treated with oral anti-CD3 mAb was significantly suppressed compared with either the PBS group or the control IgG group (P < 0.009). For cytokines, in comparison to both the PBS group and the control IgG group, IL-2 (P < 0.04) and IFN-γ (P < 0.008) secretions were significantly decreased and TGF-β secretion (P < 0.01) was significantly increased in the spleens of mice fed with anti-CD3 mAb.

We previously found that oral anti-CD3 mAb accumulates in the gut villi of mice after feeding as measured by immunohistochemistry (16). To determine whether we could recover biologically active anti-CD3 mAb after feeding, mice were fed 50 μg anti-CD3 mAb or isotype control antibody, and 6 h after feeding, anti-CD3 mAb or isotype control antibody was extracted from the upper intestine using a protein-G column. The biologic activity of the eluted anti-CD3 mAb was tested by measuring its ability to stimulate T-cell proliferation and compared with known concentrations of anti-CD3 mAb. As shown in Fig. 3, biologically active anti-CD3 mAb could be eluted from oral anti-CD3–fed animals as measured by proliferation and was equivalent to ∼0.01 μg/ml of anti-CD3 mAb.

To address whether any cell populations with regulatory function were induced after oral anti-CD3 mAb, MLN, spleen, and PLN cells from each group were prepared on day 6. Cells were stained with a combination of anti-CD4 mAb and antibody to CD25, LAP, CD45RB, and CD62L. The frequency of CD25⁺CD4⁺ cells did not change in MLN, spleen, or PLN in any groups (Table 2), whereas the frequency of LAP⁺CD4⁺ cells in spleen and PLN significantly increased in mice treated with oral anti-CD3 mAb compared with either the PBS group or the control IgG group. The frequency returned to control levels by day 21. No significant changes in frequency of either CD45RB^lowCD4⁺ or CD62L⁺CD4⁺ were observed in MLN, spleen, and PLN after oral anti-CD3 mAb (not shown), and, as in our studies of oral anti-CD3 mAb in the EAE model, we did not find that CD3 expression was decreased and/or downregulated after oral anti-CD3 mAb treatment (16).

To investigate whether the in vitro suppressive function of LAP⁺CD4⁺ cells was enhanced after oral anti-CD3 mAb, purified LAP⁺CD4⁺ cells and LAP⁻CD25⁺CD4⁺ cells from spleen on day 6 were cultured with LAP⁻CD25⁻CD4⁺ cells as responder cells, APCs, and soluble 1 μg/ml anti-CD3 mAb. As shown in Fig. 4A and B, LAP⁺CD4⁺ cells, LAP⁻CD25⁺CD4⁺ cells, and LAP⁺CD25⁻CD4⁺ cells were anergic to anti-CD3 mAb stimulation in vitro, and they suppressed responder cell proliferation in a dose-dependent fashion. The suppressive function of LAP⁺CD4⁺ cells was augmented more than two to threefold after oral anti-CD3 mAb treatment compared with that of control IgG-treated mice. However, the suppressive function of LAP⁻CD25⁺CD4⁺ cells was not affected by anti-CD3 mAb treatment. To rule out a possible effect of CD25⁺ cells on LAP⁺CD4⁺ cell suppressive function, CD25⁺ cells were excluded from LAP⁺CD4⁺ cells. As shown in Fig. 4B, LAP⁺CD25⁻CD4⁺ cells also suppressed responder cells with increased suppressive function after oral anti-CD3 mAb treatment. We next examined in vitro kinetics of suppressive function of LAP⁺CD4⁺ cells after anti-CD3 mAb feeding. LAP⁺CD4⁺ cells from mice fed with anti-CD3 mAb had enhanced suppressive function compared with LAP⁺CD4⁺ cells from mice fed with IgG on days 6 and 14, which was no longer observed on day 21 (Fig. 4C).

To determine whether the in vitro suppressive function of LAP⁺CD25⁻CD4⁺ cells after oral anti-CD3 mAb is mediated through IL-10 or TGF-β neutralizing, anti–IL-10 Ab or anti–TGF-β Ab was added in vitro. As shown in Fig. 4D, neutralizing anti–IL-10 Ab had no effect, whereas neutralizing anti–TGF-β Ab partially reversed the suppressive function by LAP⁺CD25⁻CD4⁺ cells.

To determine whether TGF-β is involved in preventing diabetes progression after oral anti-CD3 mAb in vivo, 500 μg neutralizing anti–TGF-β Ab was given on days 0, 2, 4, 6, and 14 and blood glucose levels measured on day 21. As shown in Table 3, when mice treated with oral anti-CD3 mAb were injected with neutralizing anti–TGF-β Ab, the preventive effect on diabetes progression was abrogated.

We investigated whether the suppressive function of oral anti-CD3 mAb could be adoptively transferred to naive recipients. To investigate this, CD4⁺ T-cells or CD4⁺ cells depleted of LAP⁺ cells from spleen and MLN of mice fed with anti-CD3 mAb or control IgG were transferred to naive AKR/J mice on day 0. STZ was injected to the recipient AKR/J mice on days 1–5, and blood glucose was monitored once weekly. As shown in Fig. 5, development of diabetes was significantly suppressed in the mice transferred with CD4⁺ T-cells from mice fed anti-CD3 mAb compared with the mice transferred with CD4⁺ T-cells from mice fed control IgG. No suppression was observed when CD4⁺LAP⁻ cells were transferred. Of note, depletion of LAP⁺ cells from control Ig–fed animals also enhanced the incidence of diabetes. These results confirm that LAP⁺CD4⁺ cells play a crucial role in prevention of autoimmune diseases followed by restoration of peripheral immune tolerance and that the regulatory function of LAP⁺CD4⁺ cells was reinforced by anti-CD3 mAb feeding.

In the present study we demonstrated the effectiveness of oral anti-CD3 mAb in the STZ model of diabetes. Oral anti-CD3 mAb induces an LAP⁺ regulatory T-cell that can adoptively transfer protection and functions in a TGF-β–dependent fashion. We do not know at this time whether LAP⁺ cells are a subset of T-helper (Th)3 cells, although our working hypothesis is that they are precursors of Th3 cells (21) and either differentiate into Th3 cells upon secretion of TGF-β or induce regulatory T-cells by the secretion of TGF-β. Once Th3 cells are stimulated, they may lose LAP on their surface.

Our results are analogous to what we observed with administration of oral anti-CD3 mAb in the mouse model of EAE (16). One of the features of the induction of regulatory T-cells by oral anti-CD3 mAb is a dose response in which lower doses are more effective. This was seen in the EAE model in which 5 μg was the most effective dose, no effect was observed at 50 μg, and worsening of EAE occurred with 500 μg. This dose response correlated with the induction of regulatory cells in the MLN. Thus, it appears that the dose of anti-CD3 mAb is very important for the induction of regulatory T-cells and that LAP⁺ cells will not be induced if the signal delivered to the T-cell receptor by anti–CD-3 is either too weak or too strong. We observed a similar dose-response pattern in the STZ model although the most effective dose was 50 μg, and we also observed positive effects at 250 μg. We did not test doses higher than 250 μg. The shifted dose response may be related to different strains being used and the use of complete Freund's adjuvant for the induction of EAE versus the use of STZ to induce diabetes. Of note, at 42 days after treatment of animals with 50 μg oral anti-CD3 mAb, there was no statistically significant difference between the treated and control groups although there were fewer diabetic animals in the treated group (8 of 12) than in the control group (10 of 12). For a more prolonged effect, repeat or chronic dosing may be required.

Unlike the EAE model in which oral anti-CD3 mAb enhanced recovery in animals treated at the peak of disease, we did not observe a protective effect of oral anti-CD3 mAb once animals were diabetic. The lack of effect may be related to the fact that the STZ diabetes model is an induced model with rapid onset. The NOD model of diabetes serves as one of the primary models for human disease, and we are currently testing the effect of oral anti-CD3 mAb in the NOD model given both as prevention and therapeutically at the onset of diabetes. In preliminary experiments we have found that oral anti-CD3 mAb is effective in preventing diabetes when given to neonatal NOD mice (R.M., M. Abraham, unpublished observations). Although the cause of type 1 diabetes is not known, there is the suggestion that the gut may play a role as the site at which newborns are sensitized to islet antigens related to cross-reactivity with cow milk and/or defective mechanisms of mucosal tolerance (22–24). Oral anti-CD3 mAb may thus serve to boost the natural induction of regulatory T-cells at the gut mucosa.

There is a difference in the effect of anti-CD3 mAb depending on the route it is given (IV versus oral) and when it is administered. In the EAE model, IV anti-CD3 mAb is effective when it is given at the height of the disease but not before disease induction (9). This phenomenon is felt to be related to the mechanism of action of IV anti-CD3 mAb. It is believed that IV anti-CD3 mAb kills Th1 cells that have been generated and are present at the height of disease. Oral anti-CD3 mAb, on the other hand, is effective before disease onset in the EAE model as it directly induces regulatory T-cells, which prevent the induction of effector cells. In the EAE model, it is also effective at the height of disease, presumably by boosting a naturally occurring regulatory T-cell response that occurs before the recovery phase (20). The effect of oral anti-CD3 mAb appears to be different in the STZ-induced diabetes model. We postulate that unlike EAE, regulatory T-cells have not been generated in the STZ-induced diabetes model and that the induction of regulatory cells would not be effective in later stages of STZ-induced diabetes unless Th1 responses are also downregulated. Consistent with this hypothesis, in the NOD model of diabetes, IV anti-CD3 mAb appears to eliminate the Th1 milieu, allowing the emergence of regulatory T-cells at a late time point. Unlike IV anti-CD3 mAb, oral anti-CD3 mAb directly induces regulatory T-cells; thus, in diabetes it is effective prophylactically and may also be effective after IV anti-CD3 mAb to boost a naturally occurring T-cell response that arises after IV anti-CD3 mAb (25).

After oral anti-CD3 mAb, LAP⁺ regulatory T-cells are first induced in the MLN after which they migrate to sites of inflammation (16). Thus, we observed increased numbers of LAP⁺ T-cells in the popliteal lymph nodes in EAE animals after immunization with protedlipid protein/CFA. In the present study we observed an increase of LAP⁺ T-cells in the PLNs of mice with STZ-induced diabetes. It thus appears that one of the properties of LAP⁺ regulatory T-cells is their ability to migrate to sites of inflammation where they exert their regulatory effects. In our studies of anti-CD3 mAb in the EAE model (16), we found that anti-CD3 mAb appears in the villi of fed mice within 1 hour after feeding. We now present data showing that biologically active anti-CD3 mAb can be recovered from the intestine as measured by the ability of recovered antibody to stimulate naive T-cells in vitro.

The immunological effects of oral anti-CD3 mAb are antigen nonspecific because anti-CD3 mAb binds to CD3 on the surface of T-cells independent of the antigen specificity of the T-cell receptor. Thus, it would be expected to be effective in other models of autoimmunity besides EAE. The present study is the first demonstration that oral anti-CD3 mAb is effective in an autoimmune model different from EAE and suggests that oral anti-CD3 mAb may be applicable for the treatment of human type 1 diabetes. For clinical application, it may be more effective when given to subjects at risk rather than to those with new-onset diabetes or after treatment with IV anti-CD3 mAb in subjects with new-onset diabetes.