CTLA4-Ig Restores Rejection of MHC Class-II Mismatched Allografts by Disabling IL-2-Expanded Regulatory T Cells

Introduction

Treg expansion represents a promising immunotherapy to induce functional tolerance in clinical settings (1–5). Previous work has demonstrated that in vivo Treg expansion in mice can be achieved by treatment with interleukin (IL)-2 in complex with a particular anti-IL-2 monoclonal antibody (6). In line with this, two recent clinical trials reported the efficacy of low dose IL-2 treatment to expand Tregs in refractory graft versus host disease (7) and in HCV-induced vasculitis (8), paving the way for novel treatments of several human immunological disorders such as autoimmune diseases and acute allograft rejection. Immune interventions aiming at blocking effector T cell responses through costimulatory blockade also represent efficient strategies to inhibit unwanted immune responses in vivo (9–12). This was recently suggested by the capacity of CTLA4-Ig compared to cyclosporine in improving renal function in kidney-transplanted patients (13–15). Transplantation tolerance most probably relies on a delicate equilibrium between regulatory and effector mechanisms. This raises the possibility that the combination of treatments aiming at inhibiting effector function while expanding Tregs numbers or function may represent valuable strategies to achieve immune tolerance. To test this hypothesis, we studied the effect of a therapeutic combined strategy using CTLA4-Ig and an IL-2-induced Treg expansion on allograft survival. In this setting, we demonstrate that CTLA4-Ig prevents graft acceptance induced by exogenous IL-2 therapy through inhibition of Treg homeostasis and suppressive capacities.

Materials and Methods

Treatment with IL-2/anti-IL-2 complex, CTLA4-Ig and antibodies

IL-2/anti-IL-2 complex was formed by incubating 1 μg recombinant mouse IL-2 and 9 μg of Functional Grade Purified anti-mouse IL-2 (Clone: JES6–1A12, eBioscience) for 30 min at 37°C. The complex was administered at days 0, 1 and 2 posttransplantation of each experiment. Concomitant treatment with CTLA4-Ig consists of an intraperitoneal injection of 1 mg abatacept (Orencia®, Bristol-Myers Squibb Eeig, braine l'alleud, Belgium) on days 0, 1 and 2, except for the experiment shown in Figures 4(A) and (B) where CTLA4-Ig was administered on days 6, 7 and 8. For control antibody, CTLA4-Ig was compared to purified human total IgG (>96% purity, Sandoglobuline®). Anti-CD28 mAb (Clone: 37.51, bioXcell [West Lebanon, NH, USA]; 500 μg) was administrated intraperitoneally on days 0, 1 and 2. Anti-ICOS mAb (Clone: 7E.17G9, bioXcell; 250 μg) was administrated intraperitoneally on days –1 and 1. Purified Gamma globulin from Syrian hamster (Jackson Immunoresearch Laboratories) and purified rat IgG mAb (LODNP, LO-IMEX, Université Catholique de Louvain, Brussels, Belgium) were used as control antibody for anti-CD28 and anti-ICOS mAbs, respectively.

Mixed lymphocyte culture, in vitro suppression assay and cytokine detection

For mixed lymphocyte cultures, responder cells (2.5 × 10⁶ cells/mL) isolated from draining lymph nodes (inguinal and axillary) were stimulated with irradiated splenocytes (2000 Rad; 2.5 × 10⁶ cells/mL) from either B6 (syngeneic), bm12 (allogeneic) mice in 48-well, flat-bottom plates. Cultures were incubated in 5% fetal calf serum-supplemented complete RPMI 1640 medium. For in vitro suppression assays, CD90.1 CD4⁺ T cells were isolated using a CD4 negative isolation kit (Miltenyi), then labeled with 1 μM CFSE for 30 min at 37°C and used as responder cells. CD90.2 Foxp3-GFP⁺ CD4⁺ cells were isolated with a MOFLO cell sorter and used as suppressor cells. Responders were used at a fixed concentration of 10⁵ cells per well and stimulated with 2 μg/mL of soluble anti-CD3 antibody in the presence of 4 × 10⁵ RAG^−/− spleen cells used as feeder cells. Cells were stimulated in 96-well, flat-bottom plates in triplicate. IFN-γ in culture supernatants was detected by ELISA at 48 h for mixed lymphocyte reaction (Figure 1D) and 72 h for the suppression assays, according to the manufacturer's instructions (R&D, DUOSET IFN-γ detection kit). The detection threshold was <30 pg/mL.

Results

CD28 signaling on Treg is critical for their IL-2-induced in vivo expansion and immunosuppressive capacity

We next determined whether CTLA4-Ig could inhibit exogenous IL-2-induced Treg expansion and/or suppressive capacities. Treg expansion was evaluated using Foxp3-GFP reporter ungrafted mice. Coadministration of CTLA4-Ig led to a significant reduction of peripheral Tregs in both peripheral blood (in particular at days 6 and 9 postIL-2/anti-IL-2 administration), (Figure 2A) and spleen at day 5 (Figure 2B). Of note, this negative effect on Tregs was not observed when purified human IgG, used as control antibody, were coinjected with the IL-2 complexes. However, we cannot exclude that both proliferation and survival of Tregs might be altered under costimulatory blockade. As shown in Figure (S2), an enhancement of Treg-intranuclear Ki-67 expression reflecting a strong mitotic activity was observed in complex-treated mice in comparison to untreated mice. In contrast, Ki-67 expression on Tregs was severely decreased upon CTLA4-Ig and IL-2/anti-IL-2 coinjection. In parallels, the study of Bim expression among expanded-Tregs was performed with regard to the effect of CTLA4-Ig on their survival. Indeed, Bim has been described as a typical Treg-associated proapoptotic molecule (20). We detected a smaller percentage of Bim⁺ Tregs in the IL-2/anti-IL-2-treated animals while the CTLA4-Ig cotreatment prevented this downregulation (Figure S2). The previous observations are compatible with published reports highlighting the requirement of CD28 signaling for Treg homeostasis (21–23). Accordingly, in vivo treatment with an agonist anti-CD28 mAb counteracted the inhibitory effect of CTLA4-Ig on IL-2-driven Treg expansion (Figures 2A and B). Of note, the injection of this anti-CD28 mAb alone (without exogenous IL-2) was not sufficient to expand Tregs (data not shown). A phenotypical characterization of Tregs was performed in IL-2/anti-IL-2-treated mice coinjected or not with CTLA4-Ig and anti-CD28 mAb, using Tregs from untreated mice as a control. IL-2/anti-IL-2-expanded Tregs showed an increased expression of Foxp3, inducible costimulator molecule (ICOS), membrane bound TGF-β (mbTGF-β), CD25, CTLA4 and CD39 in comparison to control Tregs (Figures 2C and S3A). The upregulation of these markers was significantly reduced by the CTLA4-Ig administration, and restored by anti-CD28 coinjection. Similarly, ex vivo suppression assays revealed the increased regulatory function of IL-2/anti-IL-2-expanded Tregs that was inhibited back to control levels by CTLA4-Ig cotreatment. Again, coadministration of agonist anti-CD28 mAbs restored the high suppressive capacity of IL-2-treated Tregs in this setting (Figure 2D). Because ungrafted animals may not be representative of recipients under conditions of ongoing alloresponses, all of these studies over Treg expansion, phenotype and functions were repeated in bm12-grafted recipients and showed similar results (data not shown).

Potent ICOS⁺ Treg induction by IL-2 is abolished by CTLA4-Ig coadministration

The notable upregulation of ICOS, a molecule whose Treg-associated expression is correlated with higher suppressor activity (24–27), led us to examine in more details the effect of CTLA4-Ig treatment on the subset of ICOS-expressing Tregs expanded by exogenous IL-2. Administration of IL-2/anti-IL-2 complex led to a significant increase in ICOS⁺ Tregs, from less than 10% in untreated control mice, up to 40% in IL-2/anti-IL-2-treated animals, a response that was almost completely antagonized by coadministration of CTLA4-Ig (Figure 3A). Notably, while IL-2/anti-IL-2 promoted both ICOS⁺ and ICOS⁻ Treg expansion, CTLA4-Ig coadministration only abrogated the ICOS-expressing subset of Foxp3 positive cells both in spleen (Figure 3B) and in lymph nodes (data not shown). ICOS⁺ expanded Tregs expressed higher levels of mbTGF-β, CTLA4 and lower levels of Bim when compared to their ICOS⁻ counterpart expanded Tregs (Figure 3C and S3B), and displayed enhanced suppressive capacities in vitro (Figure 3D). To directly assess the putative role of ICOS on Treg-mediated suppression, bm12 skins were grafted on IL-2/anti-IL-2-treated B6 recipients coinjected, or not, with anti-ICOS antibody on days –1 and 1. At day 25 posttransplantation, 80% of recipients cotreated with anti-ICOS antibody rejected their graft, whereas no rejection was observed in the IL-2/anti-IL-2-treated group (Figure 3E), a biological response that correlated with the reduced expansion of Tregs in anti-ICOS treated mice (Figure 3F).

Discussion

Our results unravel detrimental effects of CTLA4-Ig on a Treg-based protocol for suppression of allograft rejection. Although CTLA4-Ig fusion protein has been mainly described as an efficient antirejection therapy by blocking the CD28 costimulation on naïve T cells (14,29), our features clearly raise concerns about its use in tolerance induction protocols through its impact on Treg homeostasis and suppressive functions. The importance of these CD28/B7 interactions in regulatory immune function has been previously documented by a worsening of autoimmune diabetes in CD28^−/− or B7.1/B7.2^−/– NOD mice compared with wild type NOD mice (21–23). Similarly, the administration of anti-B7.1 and anti-B7.2 blocking antibodies induced rejection of spontaneously accepted MHC class II disparate heart allografts (30). Very recently, a study investigating the impact of B7/CD28 blockade by hCTLA4Ig in various allogeneic combinations of heart allografts reported an accelerated rejection of bm12 cardiac allografts by B6 recipients (31). In this combination like in the present F1 (bm12 × B6) skin graft model, spontaneous long-term allograft survival depends on Tregs and the deleterious effect of hCTLA4-Ig was accompanied with an unfavorable effector/regulatory T cell ratio (31). This is concordant with our own data showing that IL-2-mediated Treg expansion is prevented by CTLA4-Ig because of the starvation of available B7 molecules. Although a specific deficiency of CTLA4 in Foxp3⁺ CD4⁺ T cells dramatically enhances the number of peripheral Tregs (32), CTLA4-Ig administration (by blocking B7 interactions with both CTLA4 and CD28) had the opposite effect in our experiments. This remains in agreement with the established requirement of CD28 signaling for Treg maintenance (23). In this sense, we demonstrated that restoring a specific CD28 signaling was sufficient to counteract the deleterious effect of CTLA4-Ig on expanded Tregs. Noteworthy, CTLA4 signaling has been reported to be necessary for tolerance induction of full mismatched islet and skin allografts after donor specific transfusion and anti-CD40L blockade (33,34). The blockade of the B7/CTLA4 negative costimulatory signal favors potent activation of T effector cells (Teff) resulting in allograft rejection. It is not excluded that a same phenomenon could partially explain the rejection observed in the IL-2 and CTLA4-Ig cotreated group through a B7 starvation capped by CTLA4-Ig. Therefore, a specific CD28 blockade without affecting CTLA4 signaling on T cell surfaces seems to be an effective strategy for dampening immune response without affecting Tregs as demonstrated by Poirier et al. in nonhuman primate models of heart and kidney transplantation (35).

In our experiments like in others (30), allograft acceptance results from a subtle balance between Tregs and alloreactive CD4⁺ Teff. In consequence, we conclude that CTLA4-Ig-mediated costimulatory blockade has a predominant effect on Tregs counterbalancing the Teff/Treg ratio leading to allograft rejection. This is the case after IL-2/anti-IL-2 treatment as well as in the models of spontaneous acceptation of F1 (bm12 × B6) skin graft or cardiac allografts (31). Furthermore, we observed that CTLA4-Ig alone did not delay bm12 allograft rejection by B6 recipients demonstrating that B7/CD28 is not required for rejection. Besides, by using bm12-grafted B6 foxp3-GFP recipients treated or not with CTLA4-Ig, the stimulation of CD4^posGFP^neg Teff cells isolated from draining lymph nodes with donor-typed stimulators did not show any decrease in IFN-γ production (data not shown). Overall, these results indicate that CTLA4-Ig alone does not adequately control CD4⁺ mediated anti-bm12 alloreactivity.

Our results seem to be in contradiction with previous reports demonstrating a beneficial effect of CTLA4-Ig in experimental and clinical studies (9,14,15,36). First, the absence of CD8 alloreactivity clearly weakens CTLA4-Ig beneficial effects. Indeed, a previous study reported prolonged cardiac allograft survival by CTLA4-Ig in a full MHC class I plus II BALB/c to B6 combination involving CD8 T cell alloreactivity (31). In our experiments, the single MHC class II disparity rules out CD8⁺ T cell alloresponses. Moreover, the restricted disparity to three amino acids in one of the I-Ab chains is likely to render this model prone to regulation and more sensitive to a deleterious effect of CTLA4-Ig on Treg-mediated suppression of allograft rejection. Second, a major difference regarding clinical reports relies on the fact that patients always receive a combination of immunosuppressive drugs, most probably minimizing the effect of CTLA4-Ig on Tregs in terms of allograft outcome. Nevertheless, if tolerance induction protocols are considered, this aspect becomes critical because patients are weaned off immunosuppression. It is not excluded that the higher frequency of acute rejection episodes observed in the CTLA4-Ig arm of the Benefit study (14) somehow reflects Treg dysfunction. This hypothesis is supported by the reduced levels of Foxp3 mRNA detected in renal biopsies of these patients 12 months after kidney transplantation (13). A potential limitation to translate our findings in clinical transplantation lies in the formulation of the CTLA4-Ig. Indeed, we used the Abatacept form of CTLA4-Ig whereas studies in kidney-transplanted patients used Belatacept, a CTLA4-Ig with higher affinity for B7 molecules.

In the absence of CTLA4-Ig, IL-2/anti-IL-2 treatment enhanced the expression of ICOS on Tregs allowing us to distinguish two subsets of expanded Tregs: ICOS⁻ and ICOS⁺ cells. In our hands like in other studies, ICOS⁺ Tregs exhibited higher suppressive capacities than ICOS⁻ Tregs (24–26,37). Whether IL-2/anti-IL-2 complex induces ICOS expression on preexistent Tregs or expand preferentially preexistent ICOS⁺ Tregs remains to be determined. Importantly, the percentage of ICOS⁺ Tregs (not ICOS⁻) dropped after CTLA4-Ig perhaps suggesting that CTLA4-Ig prevents allograft acceptance by inhibiting only the induction of potent ICOS⁺ Tregs. In our model of Treg-mediated allograft acceptance, ICOS blockade restored rejection. This suggests that ICOS signaling plays an important role in Treg suppressive function during alloreactivity.

We conclude that IL-2 and CD28 synergize for Treg expansion, ICOS expression and the strengthening of suppressive capacities. Alloreactivity and skin graft rejection reappeared by starving Tregs of B7-CD28 interactions through the CTLA4-Ig administration. Importantly, the effect of CTLA4-Ig on both Treg expansion and suppressive function was restored by CD28 costimulation. This also raises potential issues about molecules that specifically target CD28 even if they leave CTLA4 intact (33,35). Not only did we find that CTLA4-Ig counteracted Treg expansion, it did not control CD4⁺ effector T cells in this system. Although CTLA4-Ig is unequivocally considered an efficient and nontoxic immunosuppressive agent for naive T cells, our results raise concerns about its use in protocols of tolerance induction based on Treg immunotherapy.

Disclosure

B.V. and L.M.C. are Research Fellows of the Société Francophone de Transplantation and the Fonds National de la Recherche Scientifique (FNRS-Belgium), P.L. is a Research Fellow of the Fonds Erasme (Erasme Hospital, Belgium). K.F. is an Associate Professor, Bucknell University. O.L. and A.L.M. are Research Associate of the FNRS-Belgium.

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.