CD28/CD154 blockade prevents autoimmune diabetes by inducing nondeletional tolerance after effector t-cell inhibition and regulatory T-cell expansion

Abstract

Objective: Blocking T-cell signaling is an effective means to prevent autoimmunity and allograft rejection in many animal models, yet the clinical translation of many of these approaches has not resulted in the success witnessed in experimental systems. Improved understanding of these approaches may assist in developing safe and effective means to treat disorders such as autoimmune diabetes.

Research design and methods: We studied the effect of anti-CD154 and CTLA4-Ig on diabetes development, and the requirements to induce tolerance in nod.scid mice after transfer of transgenic beta-cell reactive BDC2.5.NOD T-cells.

Results: Nod.scid recipients of diabetogenic BDC2.5.NOD cells were protected indefinitely from diabetes by a short course of combined costimulation blockade, despite the continued diabetogenic potential of their T-cells. The presence of pathogenic T-cells in the absence of disease indicates peripheral immune tolerance. T-cell maturation occurred in protected recipients, yet costimulation blockade temporarily blunted early T-cell proliferation in draining pancreatic nodes. Tolerance required preexisting regulatory T-cells (Tregs), and protected recipients had greater numbers of Tregs than diabetic recipients. Diabetes protection was successful in the presence of homeostatic expansion and high T-cell precursor frequency, both obstacles to tolerance induction in other models of antigen-specific immunity.

Conclusions: Immunotherapies that selectively suppress effector T-cells while permitting the development of natural regulatory mechanisms may have a unique role in establishing targeted long-standing immune protection and peripheral tolerance. Understanding the mechanism of these approaches may assist in the design and use of therapies for human conditions, such as type 1 diabetes.

FIG. 1.

Costimulation blockade prevents diabetes after adoptive transfer of BDC.NOD cells to nod.scid mice. A: The phenotype of cells from BDC.NOD donors, demonstrating the presence of T- and B-cells, with uniform population of vβ 4⁺ transgenic T-cells. nod.scid (N/S) mice, which lack T- or B-cells, and NOD mice are shown for comparison. B: After adoptive transfer of 2.5 × 10⁶ cells from nondiabetic BDC.NOD mice, nod.scid recipients become diabetic usually within 2 weeks. Therapy with five doses of 200 μg MR1 and CTLA4-Ig beginning the day of adoptive transfer (day 0) protects most recipients from diabetes; whereas protection was not as complete or lasting with the individual reagents, and there was no impact of this therapy on diabetes if combined treatment was delayed to day +3 after adoptive transfer.

FIG. 2.

Islet of recipients of diabetogenic cells and costimulation blockade lack inflammation. nod.scid mice were adoptively transferred without or with costimulation blockade as above. Recipients were killed at days 3, 7, 10, and 14 after adoptive transfer. Pancreata from killed mice were frozen in OCT compound in liquid nitrogen, cut, and examined using immunohistochemistry for insulin (A), CD4 (B), vβ 4 (C), and B220 (D). Recipients of cells only show progressive islet destruction after insulitis within 2 weeks after adoptive transfer. Recipients of BDC2.5 cells and costimulation blockade remain free from peri-islet inflammation and retain islets. For comparison, stained sections of pancreata from nondiabetic NOD, BDC.NOD, and nod.scid mice are shown. Shown are representative experiments of at least three independent experiments. (Please see http://dx.doi.org/10.2337/db07-1712 for a high-quality digital representation of this figure.)

FIG. 2.

Islet of recipients of diabetogenic cells and costimulation blockade lack inflammation. nod.scid mice were adoptively transferred without or with costimulation blockade as above. Recipients were killed at days 3, 7, 10, and 14 after adoptive transfer. Pancreata from killed mice were frozen in OCT compound in liquid nitrogen, cut, and examined using immunohistochemistry for insulin (A), CD4 (B), vβ 4 (C), and B220 (D). Recipients of cells only show progressive islet destruction after insulitis within 2 weeks after adoptive transfer. Recipients of BDC2.5 cells and costimulation blockade remain free from peri-islet inflammation and retain islets. For comparison, stained sections of pancreata from nondiabetic NOD, BDC.NOD, and nod.scid mice are shown. Shown are representative experiments of at least three independent experiments. (Please see http://dx.doi.org/10.2337/db07-1712 for a high-quality digital representation of this figure.)

FIG. 3.

Mice protected from diabetes after adoptive transfer and costimulation blockade retain transgenic T-cells that acquire mature memory cell markers. nod.scid mice were adoptively transferred with BDC2.5 cells and rendered diabetic and protected from diabetes using CD28/CD154 blockade as above. After 6 weeks, mice were killed, and pancreata and lymphoid organs were harvested. Pancreata were evaluated by insulin immunohistochemistry (A) in recipients for CD3⁺ and vβ 4⁺ cells via flow cytometry (B). T-cells from long-term diabetic and protected mice, BDC.NOD donors, and NOD mice were evaluated for CD62L (C) and CD44 (D). Shown are results of two separate experiments.

FIG. 4.

Proinflammatory cytokine expression in donor, diabetic, and tolerant adoptive transfer recipients. Lymphocytes were isolated from Nod.scid mice rendered diabetic, protected from diabetes with costimulation blockade (both 6 weeks after adoptive transfer), BDC.NOD mice, and NOD mice. After a 6 h in vitro incubation with 10 μg/ml BDC2.5 stimulatory peptide (RTRPLWVRME; 1040-63; 28), cells were stained for surface molecules fixed; permeabilized; stained for IL-2, TNF-α, and IFN-γ; and then evaluated by flow cytometry. During analysis, cells were gated on CD3⁺CD4⁺, which in adoptive transfer recipients usually is composed of >90–95% vβ 4⁺ cells. A representative sample of flow plots from three to four separate experiments is shown (A). Results are displayed in graphical form (B), where the P value of * vs. ** and * vs. *** is <0.05, and the P value of ** vs. *** is NS.

FIG. 5.

Mice rendered tolerant to diabetes are further protected from new diabetogenic T-cells and harbor pathogenic T-cells. A: After 6 weeks, nod.scid recipients of BDC2.5 cells protected from diabetes with costimulation blockade were given a second adoptive transfer of BDC2.5 cells in the absence of any further treatment. Unmanipulated nod.scid mice were also treated with BDC2.5 cells only. B: In other experiments, mice were protected from diabetes and others were rendered diabetic and maintained on insulin for 6 weeks. Splenocytes and LN cells were isolated from such mice and 2.5 × 10⁶ resultant cells were adoptively transferred into new nod.scid recipients. C: Lastly, mice that were rendered tolerant were treated with agents known to precipitate diabetes in other susceptible mouse models, including anti-CD25, anti-ICOS, anti-PD1, anti-TGFβ, or cyclophosphamide.

FIG. 6.

More FoxP3⁺ T-cells are found in tolerant than diabetic or donor BDC.NOD mice. Lymphocytes were isolated from pancreatic (Panc LN) and mesenteric and cervical nondraining LNs (ND LN) from nod.scid mice rendered diabetic, protected from diabetes with costimulation blockade (both 6 weeks after adoptive transfer), from BDC.NOD mice, and from NOD mice. Cells were stained for surface molecules, fixed, permeabilized, stained for intracellular FoxP3, and then analyzed by flow cytometry. FoxP3 expression gated on CD4⁺CD3⁺ cells is shown and is representative of at least four independent experiments.

FIG. 7.

Tregs are required for costimulation blockade–mediated peripheral tolerance. Donor BDC2.5 cells were depleted of CD25⁺ cells using a negative selection antibody column. Shown are representative plots of FoxP3⁺CD25⁺ expression on CD4 T-cells before (A) and after (B) depletion. C: In the absence or presence of 5 days of combined costimulation blockade therapy, 2.5 × 10⁶ of the CD25-depeleted cells were adoptively transferred into nod.scid mice, and hyperglycemia was assessed.

FIG. 8.

Anti-CD154 and CTLA4-Ig blunts early T-cell proliferation in draining pancreatic lymph nodes. BDC2.5 cells were labeled with CFSE and adoptive transferred into nod.scid mice. Some recipients were otherwise untreated, whereas others were given combined costimulation blockade on the day of adoptive transfer (day 0) and on day +2. On day +3, nondraining LNs (ND LN) and draining pancreatic LN (Panc LN) were harvested. Shown are representative flow plots of CSFE content on CD4⁺ vβ 4⁺ cells. The data are representative of three independent experiments.

FIG. 9.

Costimulation blockade induced peripheral immune tolerance by combining therapeutic and endogenous immunomodulatory mechanisms. Using these data in this report, we have developed a working model to explain our findings: After adoptive transfer of a T-cell inoculum containing Teffs and Tregs from nondiabetic BDC.NOD mice (A) into nod.scid mice, diabetogenic T-cells become activated and proliferate and destroy β-cells, which results in diabetes (A). Implicit in this model is the preferential expansion of diabetogenic Teffs over Tregs (B). In recipients treated with CD154/CD28 blockade, the expansion of diabetogenic Teffs is blunted directly by this therapy (step 1) with concomitant expansion of protective Tregs (step 2) to protect β-cells. The immunomodulatory effects of both steps 1 and 2 are required for generating and maintaining long-term peripheral tolerance by costimulation blockade.