Manipulating the immune system to induce immune tolerance has been the focus of numerous studies in transplantation aiming to promote long-term graft survival and to minimize immunosuppressive drugs with their attendant unwanted side effects. Lessons gathered from experimental models using monoclonal antibodies (Abs), or fusion proteins directed against T cell receptor (TCR) coreceptors (CD4 and CD8) and/or costimulatory molecules (CD28 and CD40L), have unanimously highlighted the central role of CD4⁺Foxp3⁺ regulatory T cells (Tregs) in controlling alloreactive effector T cells (Teff) and thus promoting transplant tolerance and long-term graft survival (1–3). Tregs, operating through mechanisms of linked suppression (tolerance to one set of antigens extended to other antigens coexpressed in the same tissue) and infectious tolerance (generation of new peripherally derived Tregs (pTreg) specific for the original antigen), are thought to be required at both the induction and maintenance phases of tolerance (4–6). Transgenic models of transplantation using CD4 and CD8 Abs have emphasized the induction of Foxp3⁺ pTreg, which can actively and continuously control Teff within the graft microenvironment, creating a form of immune privilege (6–8).

We previously established, in experimental models aiming to reverse autoimmune disease and enable acceptance of transplants, that CD3 Abs possess tolerance promoting properties (9). In particular, we reported that non-Fc-binding CD3 F(ab′)₂ fragments, administered in a defined therapeutic window, induced long-term survival of fully MHC-mismatched cardiac and pancreatic islet transplantation and implantation of allogeneic stem cells (10–12). Long-term acceptance of second grafts from the original, but not third-party donors, proved that antigen-specific tolerance had been induced (10). We have previously shown that alloreactive CD8⁺ T cells become anergic within anti-CD3 tolerized mice and use TGFβ and coinhibitory signaling as their contribution to the tolerance process (13). Our previous work revealed that Foxp3⁺ Tregs accumulated within the grafts of CD3 Ab-treated mice. However, the role of Tregs in the induction and/or the longevity of tolerance was not investigated. Here, we now show that CD4⁺Foxp3⁺ Tregs, spared from depletion by CD3 Ab, are crucial for the induction of transplant tolerance, whereas other alloantigen-specific CD4⁺ T cells become anergic, as was the case for CD8⁺ T cells (13). We conclude that Treg may be relatively dispensable for tolerance maintenance, which predominantly manifests itself as T cell anergy.

A large body of experimental work in autoimmunity and transplantation has implicated Foxp3⁺ Tregs in orchestrating and sustaining dominant mechanisms of tolerance, allowing a constant control of Teff. Data presented in this article support the fact that transplant tolerance relies on cell extrinsic and intrinsic mechanisms engaging both FoxP3⁺ Tregs and CD4⁺ T cell anergy in a timely manner. By using a fully mismatched islet allograft model and CD3 Ab therapy, we demonstrated that CTLA-4⁺Foxp3⁺ Tregs resist depletion by the CD3 Ab, unlike effector alloreactive Gzmb⁺Tbx21⁺CD4⁺ Th1 T cells, so enabling them to orchestrate the subsequent tolerance inducing events. Even though FoxP3⁺ Treg could be found in graft infiltrates indefinitely, treatment of hosts maintaining their grafts long term with the PC61 and CTLA-4 Abs did not break the tolerant state. Detailed analysis of graft-infiltrating T cells was consistent with the alloreactive members being anergic.

Our previous work demonstrated the capacity of CD3 Ab to induce transplant tolerance (10–12). The therapeutic window was of crucial importance, and transplant tolerance was only obtained when the treatment was applied at the time of Teff priming to the alloantigens. One direct effect of the Ab was a partial and transient depletion of CD4⁺ and CD8⁺ T cells in secondary lymphoid organs and in the grafts. Importantly, all T cell subsets were not equally sensitive to the CD3 Ab-depleting effect. Single-cell multiplex PCR, optimized to discriminate T cell heterogeneity and functional behavior (19, 20), showed that CD3 Ab preferentially targeted and eliminated CD4⁺ T cells coexpressing Gzmb/Tbx21. This finding is reminiscent of our previous report showing that cytotoxic intragraft CD8⁺ T cells coexpressing granzyme B and perforin were selectively deleted (13). Apoptosis of these alloreactive T cells by CD3 Abs was mandatory for long-term graft survival and was dependent on the Fas/FasLigand pathway: in vivo FasL blockade abrogated T cell elimination and CD3 Ab-induced tolerance to islet transplant (13). In this study, we demonstrated that in vivo, in contrast to Teff, Foxp3⁺ Tregs were resistant to CD3 Ab-mediated depletion. Consequently, their proportion increased in both lymphoid organs and the transplanted islets at the end of treatment. The molecular mechanisms underlying such Treg resistance to apoptosis are not fully understood, but may include a reduced expression and activation of signaling molecules downstream of the TCR/CD3 complex as well as a decreased expression of the CD3 molecules at the cell surface compared to CD4⁺Foxp3⁻ T lymphocytes (21–23).

Regulatory T cells are mandatory for the induction of tolerance as their depletion at the time of CD3 Ab therapy, or for a short time thereafter, abolished long-term graft survival demonstrating a critical role of Tregs in that early period when the immune system is responding to graft alloantigens. At this time, numerically dominant Tregs may influence residual and reconstituting naïve T cells inhibiting their differentiation into Th1 effectors. During this phase, Tregs from CD3 Ab-treated mice become endowed with potent antigen-specific suppressive properties as they appeared to inhibit proliferation and IFN-γ secretion of effector cells stimulated with donor antigens more efficiently than Tregs from untreated recipients, or effector cells specific for third-party antigens. CTLA-4 was necessary for Treg-suppressive functions as CTLA-4 blockade completely abrogated their inhibitory effect in vitro and prevented CD3 Ab-induced tolerance and long-term survival in vivo.

Our results are in accordance with many experimental studies reporting that Treg depletion (mainly by using the PC61 Ab) around the time of grafting prevented the induction of transplant tolerance (Table S1 in Supplementary Material). However, very few have addressed the impact of Treg elimination once tolerance has been established. Surprisingly, in our model, tolerance was sustained after administration of PC61 or CTLA-4 Abs on day >100 posttransplant, thereby suggesting that long-term maintenance of tolerance was not solely dependent on CTLA-4⁺Foxp3⁺ Treg activity. Survival of islet allografts in Foxp3^DTR mice after DT injection further supported this point. It was possible that other types of regulatory T-cells were operating within the graft. For example, IL-10-producing T-cells have been implicated in some forms of therapeutic tolerance (24–26). We have excluded this by demonstrating that CD3 Ab therapy can tolerize even in IL-10 knockout mice (unpublished data). It may also be possible that access of long-term tolerated grafts to the inhibitory effects of Ab may not be equivalent in the inductive and long-term maintenance phases of tolerance.

However, we have provided evidence supporting the notion that tolerance maintenance is associated with the persistence of CD4⁺ anergic T cells. First, single-cell multiplex PCR showed that, following CD3 Ab-induced apoptosis of alloreactive Gzmb⁺Tbx21⁺ Th1 cells, most of the remaining CD4⁺ T cells expressed none or only one of the selected inflammatory genes over long term, thereby highlighting the absence of potent effector activity in tolerated grafts. Second, in accordance with PCR data, intragraft CD4⁺ T cells show weak proliferation capacities as revealed by the drastically reduced expression of the proliferation marker Ki67 and of the β subunit of the IL-2 and IL-15 receptors (CD122) up to day +100 posttransplant. Third, we demonstrated that total CD4⁺ T cells or Treg-deprived CD4⁺ T cells isolated from tolerant C57BL/6 mice were unable to efficiently respond to donor antigen stimulation in contrast to third-party antigens, this intrinsic functional unresponsiveness being sustained over the long term (>100 days posttransplant). Fourth, we observed that tolerated islet grafts harbor a significant proportion of CD4⁺ T-cells exhibiting a Foxp3⁻CD73^hiFR4^hi phenotype, which has been correlated with an anergic state (16–18). Fifth, intragraft CD4⁺ T cells expressed high levels of the inhibitory receptor PD-1, and our recent report showed that blocking the PD-1/PD-L1 pathway at the time of CD3 Ab therapy or at day +100 posttransplant abolished immune tolerance and permanent islet graft survival (13). In addition, we have recently demonstrated that tolerated islets from CD3 Ab-treated C57BL/6 mice harbor CD8⁺ T cells deprived of proliferative, inflammatory, or cytotoxic capacities and that CD8⁺ T cell anergy was dependent on the presence of the alloantigens and on an in situ crosstalk between the PD-1/PD-L1 and TGFβ/TGFβRII pathways (13). This possibility of wholesale T-cell anergy can itself be sufficient to explain why PC61 and anti-CTLA-4 Abs failed to break the established tolerant state. In vivo demonstration of anergy to alloantigens has been poorly documented so far. One study in rat models showed that long-term survival of renal transplant after donor blood administration was associated with the inability of T cells to produce IL-2 and to proliferate and that treatment with recombinant IL-2 abrogated tolerance (27).

Our data are in line with a recent report documenting in type 1 diabetic (T1D) patients treated with humanized CD3 monoclonal Ab (teplizumab), the presence of dysfunctional CD8⁺ T cells qualified as “exhausted-like” and expressing the transcription factor Eomes, TIGIT, and KLRG-1 (28). In our islet transplant model, CD8⁺ T cells from CD3 Ab-treated tolerant mice also expressed Eomes and may well correspond to the subset described in T1D patients treated with CD3 Abs (13). In our model, we use the term “anergy” rather than “exhaustion” for both CD4⁺ and CD8⁺ T cells to define the unresponsive subsets in CD3 Ab-treated mice. In fact, although these cells did not respond to donor antigens, they maintained efficient responses to third-party antigens or polyclonal stimulus.

In conclusion, this study provides compelling evidence that Tregs play a substantive role in establishing a state of tolerance but that other mechanisms such as T cell anergy become more prominent for its maintenance and for long-term graft survival. This does not, of course, rule out a role for Foxp3⁺ T-cells in tolerance maintenance, as the level of their activity may, through homeostatic mechanisms, have been adjusted to that corresponding to residual effector function in the treated hosts (29, 30). These finding may be of importance for the implementation of Treg cell therapy in organ transplantation (31). One key point concerns Treg viability and persistence in vivo. In a hematopoietic stem cell transplantation clinical trial, infused Tregs were no longer detectable in the circulation after 2 weeks (32). It remains unclear whether the infused Tregs had become epigenetically destabilized and lost Foxp3 expression, had migrated to tissues, or had even died. Nonetheless, our data suggest that infused Tregs might enable sufficient immediate control of Teff function around the time of transplantation and immunotherapy, so permitting diverse intrinsic tolerogenic mechanisms to emerge and deliver the maintenance of a life-long tolerant state, with anergy being one such dominating mechanism.

AB and MB designed and performed experiments, analyzed the data, and wrote parts of the manuscript. TG and FV performed experiments and analyzed the data. HW and LC provided critical discussion and advices and reviewed the manuscript. SY designed and directed the study, analyzed the data, and wrote the manuscript.

All authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.