Costimulation Blockade of Both Inducible Costimulator and CD40 Ligand Induces Dominant Tolerance to Islet Allografts and Prevents Spontaneous Autoimmune Diabetes in the NOD Mouse

Adult C57BL/6 (H-2^b), BALB/c (H-2^d), and CBA/JCr (H-2^k) mice were purchased from Charles River (St. Constant, Quebec, Canada) and housed under standard conditions. Immunodeficient C57BL/6-RAG1–knockout mice (Jackson Laboratory, Bar Harbor, ME) and 4-week-old female NOD mice (Taconic, Mississauga, ON, Canada) were housed under specific pathogen–free conditions. All animals were cared for in accordance with the guidelines established by the Canadian Council on Animal Care.

C57BL/6 and immunodeficient C57BL/6-RAG–knockout recipient mice were rendered chemically diabetic by a single injection of streptozotocin (STZ; 200 mg/kg i.v.; Sigma-Aldrich, Oakville, ON, Canada). Donor islets were isolated from fully major histocompatibility complex–mismatched BALB/c mice by collagenase digestion (1 mg/ml; Sigma-Aldrich) followed by Ficoll purification (Sigma-Aldrich) (23,24). Next ∼500 islets were transplanted under the left renal capsule of diabetic recipient mice. Allograft function was monitored by serial blood glucose measurements. Successful engraftment was defined by the correction of the serum glucose level to <8 mmol/l by the 3rd day after the transplant, and graft rejection was defined as a rise in serum glucose levels to >15 mmol/l for 2 consecutive days.

The CD40 ligand mAb (MR1) was purchased from Bio-Express (West Lebanon, NH). The production and characterization of a nondepleting anti-ICOS mAb (12A8) have been previously described (17). An isotype-matched IgG2b control mAb (Bio-Express) was used as the control for anti-ICOS mAb therapy. Islet allograft recipients were treated with CD40 ligand mAbs (0.25 mg; days 0, 2, 4, and 6) and/or anti-ICOS mAbs (0.1 mg/day × 14 days) beginning on the day of transplantation. Female NOD mice were treated with an IgG2b control mAb (0.1 mg/day × 14 days) or CD40 ligand mAb (0.25 mg; days 0, 2, 4, 6, and 10) with or without an anti-ICOS mAb (0.1 mg/day × 14 days). Treatment in NOD mice was initiated at age 10 weeks, during the development of insulitis but before the onset of spontaneous diabetes. All mAbs were given intraperitoneally.

Splenocytes were cultured in duplicate wells containing 2 × 10⁵ responder cells and dilutions of irradiated (1,500 Rad) stimulator cells. Responder cells were obtained from naïve and long-term engrafted mice, and stimulator cells were derived from C57BL/6 (syngeneic), BALB/c (donor), and CBA/J (third party) mice. After being cultured at 37°C for 3 days, the cells were pulsed for 18 h with 1 μCi ³H-labeled thymidine/well and harvested, after which the thymidine incorporation was determined.

Cytotoxic responses were assayed by the JAM test (25). Briefly described, 5 × 10⁶ spleen cells from C57BL/6 naïve and tolerant mice (responders) were stimulated for 5 days with 2 × 10⁶ irradiated BALB/c (donor) spleen cells. Con A blast targets were set up 40 h before CTL assay by culturing 1.5 × 10⁶ naïve syngeneic and BALB/c spleen cells with Con A (1.25 μg/ml), then labeling them with ³H-thymidine. The lysis of target cells was tested at various responder/target ratios.

A single-cell suspension of donor splenocytes from the spleen of BALB/c mice was passed through a nylon mesh. Viable cells were counted and adjusted to 20 × 10⁶ cells/ml. Long-term engrafted mice previously treated with anti-ICOS and CD40 ligand mAbs were immunized with 5 × 10⁶ spleen cells (0.25 ml) intraperitoneally and followed for graft rejection.

Long-term graft function in mice maintaining normoglycemia for >200 days was confirmed by a return to hyperglycemia after nephrectomy of the kidney bearing the islet graft. To test for donor-specific immunologic tolerance, nephrectomized mice underwent retransplantation of same donor strain (BALB/c) or third-party (CBA/J) islets into the remaining contralateral kidney. No immunosuppressive therapy was given, and blood glucose was monitored serially to detect graft rejection.

C57BL/6 mice treated with anti-ICOS and anti–CD40 ligand mAbs and C57BL/6-RAG–knockout immunodeficient mice underwent nephrectomy of their islet-bearing kidneys after maintaining normoglycemia for >100 days. The composite islet (BALB/c) and kidney (C57BL/6) grafts from both groups of mice were transplanted into STZ-administered, diabetic, naïve C57BL/6 mice, and the recipients’ blood glucose levels were monitored. Renal transplantation was performed as previously described (26) and involved flushing the donor kidney with cold saline immediately after nephrectomy followed by heterotopic implantation of the graft, with the donor renal artery and vein being anastomosed to the recipient aorta and inferior vena cava, respectively. An anastomosis between the donor ureter and recipient bladder was also created.

After nephrectomy of graft-bearing kidneys, one half of the islet allograft was snap frozen, and cryostat sections were stained by immunoperoxidase using mAbs to CD4 and CD25 or isotype controls (Pharmingen, San Diego, CA), rabbit anti-mouse Foxp3 or control rabbit IgG, and an Envision kit (Dako, Carpinteria, CA) (17). The rabbit antibody was generated against a 13-mer peptide TFPRSGTPRKDSN corresponding to amino acids 169–181 of mouse Foxp3 and affinity purified. Control studies showed that this antibody produced a single band of ∼52 kDa on Western blotting of mouse thymocytes and stained murine Foxp3⁺ cell transfectants but not untransfected cells. The second half of each tissue was fixed in formalin, and paraffin-embedded sections were stained with hematoxylin-eosin to assess overall cellularity.

Graft survival was analyzed using the Kaplan-Meier method, and comparisons among groups were performed by the log-rank test (SPSS, Chicago, IL).

Compared with untreated mice, anti-ICOS mAb treatment did not prolong graft survival (mean survival time 13 days, n = 11), whereas anti–CD40 ligand mAb therapy resulted in significant prolongation of graft survival, with 60% of grafts (6 of 10) surviving >100 days. Combination therapy with anti-ICOS and anti–CD40 ligand mAbs resulted in the most potent prolongation of graft survival, with 93% of grafts (26 of 28) being maintained for >200 days (Fig. 1). This level of acceptance was significantly higher than that achieved by monotherapy (P < 0.05).

Standard mixed lymphocyte responses to donor and third-party antigens (Fig. 2A) and CTL reactions to donor antigens (Fig. 2B) were assayed in mice maintaining normoglycemia for >200 days. No significant difference in proliferative or cytotoxic responses between lymphocytes from long-term engrafted mice and those from naïve mice was detected. Because mice treated with combination therapy maintained allograft function but showed intact in vitro responses to donor antigens, we considered whether grafts were maintained on the basis of immunologic ignorance. To test this, we immunized long-term engrafted mice with 5 × 10⁶ donor spleen cells as an immunologic challenge to precipitate rejection. However, all islet allografts (n = 5 of 5) continued to maintain normal function for >50 days, suggesting that long-term allograft acceptance was not simply due to immunologic ignorance.

Although immunization with donor spleen cells did not trigger allograft rejection, it remained possible that recipients were in fact “ignorant” of donor antigen but unable to reject a well-established graft that may not have delivered the necessary signals needed to recruit effector T-cells to the graft site (27). To more fully determine whether graft acceptance was due to immunologic ignorance or tolerance, we challenged long-term engrafted recipients with a second, same donor islet allograft.

Long-term allograft acceptance (>200 days) was first confirmed in mice treated with anti-ICOS and anti–CD40 ligand mAbs by nephrectomy of the graft-bearing kidney, which resulted in a prompt return of hyperglycemia in all cases (n = 11 of 11). Mice were then challenged with a second donor (BALB/c) islet allograft in the remaining contralateral kidney without further immunosuppressive therapy. All second donor allografts were accepted long term (>100 days), whereas a second cohort of tolerant mice challenged with third-party (CBA/J) islets rapidly rejected their grafts (Fig. 2C), demonstrating that donor-specific tolerance had been achieved. Because 60% of mice treated with anti–CD40 ligand mAbs alone also demonstrated long-term allograft survival, we challenged these mice with a second, same donor islet allograft and found that only two of four mice had long-term acceptance of the second graft. This suggested that in addition to increasing the proportion of mice achieving indefinite allograft survival, the combination of anti-ICOS with anti–CD40 ligand also impacts the quality of the tolerant state.

We next sought to address the issue of long-term maintenance of immunologic tolerance in vivo despite the presence of strong antidonor proliferative and cytotoxic responses in vitro. Long-term allograft acceptance in our model could have been dependent on a regulatory mechanism that can actively constrain alloaggressive T-cell responses (28). Given that splenic lymphocytes from tolerant mice generated robust proliferative responses to donor antigen in vitro (Fig. 2A), it appeared unlikely that regulatory T-cells were present in the spleens of tolerant mice. To more fully rule out this possibility, a standard mixed-lymphocyte reaction assay was used to detect the presence of donor-specific regulatory T-cells. Lymphocytes from tolerant mice did not suppress the alloresponses of cocultured lymphocytes from naïve mice (data not shown), suggesting that donor-specific regulatory T-cells are not present in the spleen of mice treated with combination therapy. However, it is possible that the specificity or effector function of these cells is restricted to islet antigen alone, rendering them undetectable in assays using splenocyte targets. Regulatory T-cells have also been identified in recent years within the grafts of tolerant mice and are capable of mediating a dominant-suppressive effect on naïve T-cells (29). Therefore, a regulatory mechanism restricted to the graft site may have been operating in long-term engrafted mice treated with anti-ICOS and anti–CD40 ligand mAbs. Immunoperoxidase staining of tolerated islet allografts showed peri-islet staining of infiltrating mononuclear cells for CD4, CD25, and Foxp3, whereas expression of these markers was not seen in healed-in allografts from untreated B6-RAG–knockout mice (Fig. 3); sections labeled with control antibodies were unstained. Although these findings do not demonstrate a functional regulatory role of the intragraft T-cells, they do suggest the possibility that tolerance of the islet graft may be dominant and dependent on recipient T-cells.

To functionally evaluate the state of the tolerated graft and the possible involvement of dominant mechanisms, we performed a retransplant procedure involving the transfer of tolerated islet grafts from long-term engrafted mice to naïve mice. Composite islet (BALB/c) and kidney (C57BL/6) grafts from tolerant mice treated with anti-ICOS and anti–CD40 ligand mAbs were explanted and retransplanted into diabetic, naïve C57BL/6 mice. Control composite islet (BALB/c) and kidney (C57BL/6) composite grafts from STZ-administered C57BL/6-RAG–knockout mice were also harvested after long-term maintenance of normoglycemia (>50 days) and retransplanted into diabetic, naïve C57BL/6 mice. Islet allografts from both long-term engrafted (>100 days) wild-type and immunodeficient recipients were well preserved before the retransplantation (Fig. 4B–D). However, although grafts from immunocompetent, treated recipients were accepted over the long term in secondary naïve recipients mice, grafts from control, immunodeficient mice were rejected (Fig. 4A and E). Moreover, the expression of CD4, CD25, and Foxp3 in donor allografts from tolerant, treated mice was maintained long term (>100 days) after graft retransplantation into naïve mice (Fig. 4F). The rejection of the control islet grafts from immunodeficient mice demonstrated that it was not the healed-in characteristic of the graft or “parking” that allowed for long-term acceptance of islet grafts from tolerant, treated mice. These results indicate that treatment with anti-ICOS and anti–CD40 ligand mAbs allowed donor islets to be accepted by a dominant mechanism involving recipient T-cells.

Successful tolerance protocols in islet transplantation must be effective at preventing alloimmune rejection and also in overcoming the underlying autoimmune process of diabetes. Therefore, having shown that combined therapy can prevent islet allograft rejection, we tested the effectiveness of this therapy in preventing autoimmune islet destruction by treating female NOD mice with anti-ICOS and anti–CD40 ligand mAbs, either alone or in combination, for 14 days, beginning at age 10 weeks. Female NOD mice in our colony develop islet infiltration by leukocytes by about age 6–7 weeks. Without immunomodulation, ∼50% of these mice develop diabetes by age 25 weeks and 75% develop diabetes by age 33 weeks. In this study, treatment of NOD mice with an IgG2b control mAb did not alter the rate of diabetes as compared with untreated mice (Fig. 5A). Monotherapy with either anti-ICOS or anti–CD40 ligand mAbs resulted in a marked but nonsignificant reduction in diabetes onset compared with mice treated with control mAbs (37 and 35 vs. 63%, respectively; P > 0.05). The combination of anti-ICOS mAbs with anti–CD40 ligand mAbs led to a more potent reduction in the onset of diabetes, with only 11% of mice (2 of 19) becoming diabetic (P = 0.065 vs. individual monotherapies; P < 0.001 vs. control mAbs) (Fig. 5B). Taken together, these results indicate that dual blockade of ICOS and CD40 ligand signaling is highly effective at preventing alloimmune rejection and autoimmune destruction of islet cells.

T-cell activation and allograft rejection can occur despite blockade of individual costimulatory molecules such as CD28, CD40 ligand, or ICOS. Furthermore, although renal allografts in nonhuman primates can survive for >1 year after discontinuation of anti–CD40 ligand treatment, islet allografts are rapidly rejected within several months (2,3), suggesting that more robust strategies involving two or more agents may be required to induce islet transplant tolerance. The results of the current study support this concept by demonstrating that the blockade of ICOS and CD40 ligand signaling is significantly more effective in facilitating islet allograft acceptance when used in combination rather than as individual monotherapies.

A rationale for combining ICOS and CD40 ligand targeting is suggested by studies showing that although anti-ICOS monotherapy provides some protection to heart and liver allografts from rejection (18,21), it is ineffective at promoting islet graft acceptance (30). In contrast, when anti-ICOS mAbs are combined with blockade of CD28-B7 signaling, FK506 therapy, or rapamycin treatment, islet allograft survival is significantly improved (19,22). In a similar fashion, although blockade of CD40 ligand signaling alone is not effective at inducing tolerance in stringent models of islet and skin allotransplantation, adjunctive strategies such as donor-specific transfusion (31), CD45 signaling blockade (32), blockade of the adhesion/homing receptor LFA-1 (33), or concurrent stimulation of negative signaling through programmed cell death 1 (34) can all lead to indefinite islet allograft acceptance.

The synergy we observed between ICOS and CD40 ligand blockade may involve complementary inhibition of CD4⁺ and CD8⁺ T-cell responses. Inhibition of CD40 ligand signaling can effectively prevent donor-specific CD4⁺ T-cell responses but has little effect on alloreactive CD8⁺ T-cells (35), which can still mediate allograft rejection (36,37). In stringent models, strategies to induce tolerance through CD40 ligand blockade require adjuncts, such as CD8 mAbs or CTLA4-Ig, that can provide direct anti-CD8⁺ T-cell activity (10,11,36). In our model, the combination of anti–CD40 ligand mAbs with anti-ICOS mAbs provided a similar beneficial effect, consistent with a key role of ICOS in regulating the expansion and differentiation of CD4 and CD8 effector cells (19,21). The benefit of this combination is of particular interest in islet transplantation given that CD8⁺ cells have been reported to be important effectors of rejection in murine models of islet transplantation (38).

After demonstrating the efficacy of combined ICOS and CD40 ligand blockade, we assessed whether tolerance had been achieved in this model. In vitro analysis of T-cells harvested from long-term engrafted mice showed potent proliferative and cytotoxic responses to donor alloantigens, indicating that clonal deletion of donor-reactive T-cells was neither achieved nor required for the long-term acceptance of islet allografts after combination therapy. To test for immunologic ignorance as a mechanism for allograft survival, immunization with donor splenocytes was performed in long-term engrafted mice, with the result that it failed to trigger allograft rejection. Because a well-established allograft may be refractory to rejection despite immunization (39), we challenged long-term engrafted mice with a second donor islet allograft to more thoroughly exclude ignorance as a mechanism and to test for immunologic tolerance. These mice accepted same donor but promptly rejected third-party allografts, indicating that anti-ICOS and anti–CD40 ligand mAb treatment induces donor-specific tolerance. One possibility for these divergent findings may be the generation of split tolerance, wherein the recipient may be tolerant to the islet allograft but not to donor splenocytes (40). Another possible explanation is that tolerance in this model is dependent on a dominant regulatory mechanism that controls peripheral alloreactivity.

Recently Graca et al. (29) demonstrated that CD4⁺ regulatory T-cells were present in tolerated skin allografts and were capable of mediating dominant transplantation tolerance. Moreover, Hori et al. (41) reported on a novel transcription factor specifically expressed by CD4⁺CD25⁺ regulatory T-cells, namely Foxp3. This gene represents a specific marker for regulatory T-cells, unlike CD25, CD45RB, and GITR, which are also expressed on activated effector or memory T-cells. More recently, Cobbold et al. (42) demonstrated that tolerant recipients of skin allografts after CD4 Ab blockade had intragraft regulatory T-cells that expressed high levels of Foxp3 mRNA. In the current study, immunohistology of tolerated islet allografts showed peri-islet staining for CD4⁺CD25⁺ T-cells and the presence of Foxp3, suggesting a possible intra-graft regulatory mechanism for the maintenance of tolerance. This hypothesis is supported by the fact that composite islet and kidney grafts from tolerant mice were accepted indefinitely when retransplanted into naïve B6 mice, whereas healed-in composite islet and kidney grafts from control B6-RAG–knockout mice devoid of CD4⁺CD25⁺Foxp3⁺ T-cells were rejected. These findings indicate that acceptance of composite grafts from treated recipients is not a consequence of reduced immunogenicity of the graft secondary to being healed in (27) and/or depleted of passenger leukocytes (43) but that treatment with anti-ICOS and anti–CD40 ligand alters the graft microenvironment in facilitating tolerance. The presence of CD4⁺CD25⁺Foxp3⁺ T-cells in the graft, together with the dependence on recipient lymphocytes for tolerance of the retransplanted islets, suggests that regulatory T-cells may have been acting on anti-donor T-cells. However, even though the treated recipients were clearly tolerant rather than ignorant, we cannot fully exclude the additional possibility that the recipient T-cells acted directly on the graft to reduce the graft’s immunogenicity. Nevertheless, these observations provide important initial data regarding a role for recipient T-cells in generating a transferable operationally tolerant state. The association between the presence of intragraft lymphocytes and the maintenance of tolerance could have been further strengthened by comparing allografts that were rejected to those that were accepted indefinitely in recipients treated with anti-ICOS and anti–CD40 ligand mAbs; however, there were too few recipient mice that actually rejected their allograft (2 of 28) for this comparison to be made. Further study is required to determine if in fact the CD4⁺CD25⁺Foxp3⁺ intragraft T-cells have a functional regulatory role.

Having demonstrated the efficacy of combined anti-ICOS and anti–CD40 ligand blockade in the regulation of alloimmune responses, we proceeded to test this strategy in the prevention of autoimmune diabetes in the NOD mouse. Consistent with previous reports (44), we found that monotherapy with the CD40 ligand reduced the onset of diabetes. In an interesting finding, the blockade of ICOS signaling alone demonstrated a superior effect over CD40 ligand blockade at delaying diabetes onset, although the incidence of diabetes was similar in both groups by age 30 weeks. Although monotherapy with either agent demonstrated efficacy, the combination of these agents resulted in a more profound reduction in diabetes. In our study, therapy was administered after the onset of insulitis but before the onset of diabetes. The timing of costimulation blockade was based on evidence that blockade of ICOS signaling during antigen priming can result in more severe disease, whereas blockade during the efferent immune response (i.e., immediately before disease onset) can prevent its progression (45).

In summary, we have demonstrated that combination treatment with anti-ICOS and anti–CD40 ligand is a potent strategy in inducing long-term islet allograft acceptance. The maintenance of donor-specific tolerance despite the presence of alloreactive T-cells includes the presence of T-cells at the site of the tolerated allograft that express the regulatory markers CD4, CD25, and Foxp3. In addition, we have shown that this combination therapy can also significantly reduce the onset of primary autoimmune diabetes in NOD mice, indicating that its effectiveness is not limited to controlling alloimmune responses. These findings underscore the efficacy of simultaneous blockade of ICOS and CD40 ligand signaling as a potential therapy in clinical islet transplantation and emphasize the need for further studies in large animal models.

S.A.N. receives salary support from the Canadian Institutes of Health Research (CIHR) and the Alberta Heritage Foundation for Medical Research. A.M.J.S. is supported by a jointly funded Wyeth–Canadian Institutes of Health Research (CIHR)/University of Alberta Clinical Research Chair in Transplantation and through a Clinical Center Grant from the Juvenile Diabetes Research Foundation. Additional support for this work was from a CIHR operating grant to C.C.A. and National Institutes of Health Grants AI-54720 and DK-63591 to W.W.H.