The Challenge of Inhibiting Alloreactive T-Cell Memory


Corresponding author: A. Valujskikh,


Memory T cells specific for donor antigens present a unique challenge in transplantation. In addition to expressing rapid and robust immune responses to a transplanted organ, memory T cells may be resistant to the effects of currently used graft-prolonging therapies. The increasing recognition that alloreactive memory T cells participate in transplant rejection is driving new lines of research focusing on understanding the immunobiology of alloreactive memory T cells and on designing novel therapies to specifically target memory T cells. The purpose of this review is to summarize the effects of existing immunosuppressive drugs and costimulatory blockade on functions of alloreactive memory T cells that undermine allograft survival.


The generation and maintenance of immunologic memory accounts for the enhanced responses upon re-encounter with the same pathogen and is a characteristic feature of adaptive immunity. When compared to microbial infections, exposure to alloantigens is less common. Nevertheless, the T-cell repertoire of many humans contains alloreactive T cells with a memory phenotype. There is accumulating evidence that T cells primed through infections, immunizations and environmental exposure may potentially cross-react with allogeneic MHC molecules (so-called ‘heterologous immunity’) (1). In addition, individuals may become sensitized to allo-MHC molecules during blood transfusions, pregnancies and rejection of a previous transplant. Finally, recent studies in rodents suggest that allospecific memory T cells are generated during homeostatic expansion of T cells in patients undergoing lymphoablative therapies (1).

Regardless of origin, it is likely that the features of alloreactive memory T cells essential for host protection become a liability in transplant settings. Due to the lower activation threshold compared to naïve T cells, reactivation of memory T cells may occur outside of the secondary lymphoid tissue and does not require interaction with professional APC (2,3). Upon activation, donor-specific memory T cells can contribute to allograft destruction through multiple mechanisms. First, both CD4 and CD8 memory T cells proliferate within hours after stimulation, giving rise to populations of effector T cells that produce proinflammatory cytokines and exhibit direct cytotoxic function (3). Memory CD4 T cells also provide efficient help for the activation of naïve T cells and antibody production that, in turn, mediate allograft destruction (4). Furthermore, initial contact of memory T cells (especially CD8 memory T cells) with donor endothelium may facilitate infiltration of the recipient's immune cells into the graft via up-regulation of adhesion molecules and chemokines (5).

Based on the current knowledge of allograft rejection mechanisms, it would be logical to assume that the extent of memory T-cell-mediated injury to the transplanted tissue is directly proportional to the numbers of donor-specific memory T cells in any given individual. According to this hypothesis, the influence of the memory T-cell population on allograft outcome will be minimal in nonmanipulated rodents housed in pathogen-free facilities (and generally assumed to be immunologically naïve) and the greatest in large animals or human individuals with an extensive history of environmental exposure. Supporting this theory, the experimental animals primed to alloantigens through allograft rejection (and containing several fold higher numbers of allospecific memory T cells than nonprimed animals) exhibit accelerated rejection of a second transplant from the same donor (‘second set rejection’). Analogously, in human transplant patients higher frequencies of donor-specific memory T cells in the peripheral blood at the time of transplantation are associated with poor allograft outcome (2). Thus, controlling the responses of memory alloreactive T cells is likely to improve allograft outcome. The intention of this review is to summarize the effects of existing immunosuppressive drugs and costimulatory blockade-based approaches on the survival and functions of alloreactive memory T cells.

Alloreactive Memory T Cells and Immunosuppression

Despite the extensive research on currently used and newly developed immunosuppressive drugs, very little is known about the direct effect of these agents on memory T-cell responses. The available information is often controversial and hard to analyze due to differences in model systems evaluating the effect of the particular drug, in sources and phenotypes of memory T cells and in functional readouts used in each study. The interpretation of such studies is further complicated by the fact that even in the absence of a direct influence on pre-existing memory T cells, immunosuppressive agents may limit responses of naïve donor-specific T cells and prevent generation of new memory T cells resulting in improved allograft outcome in comparison to nontreated sensitized recipients.

The most clinically relevant data on the relationship between memory T cells and immunosuppression come from analyses of T-cell responses in transplant patients. It has been well established that donor-reactive T cells are detected in many transplant recipients despite ongoing immunosuppressive treatment (6). The analysis of mumps-specific T-cell responses clearly demonstrates that pre-existing T-cell memory is not eliminated by a course of immunosuppression (7). These results suggest that existing immunosuppressive regimens might be ineffective in preventing the memory T-cell contribution to allograft destruction. Moreover, recent studies in primates and humans clearly show that some of the currently used lymphoablative strategies lead to the opposite outcome: preferential expansion and accumulation of T cells with a memory phenotype.

Nonetheless, successful attempts to control memory T-cell responses have been reported. The calcineurin inhibitors CsA and FK506 inhibited in vitro proliferation and cytokine production of memory CD4+CD45RA T cells derived from renal transplant patients (8). The NF-κB nuclear translocation blocker 15-deoxyspergualin (DSG) prevented activation of donor antigen-specific memory CD8 T cells and synergized with costimulatory blockade to induce long-term skin allograft survival in a mouse model (9). Administration of the sphingosine-1 phosphate receptor agonist FTY720 resulted in lymphoid sequestration of donor-specific memory CD4 T cells delaying heart allograft rejection in mice (10). Overall, these results indicate that immunosuppressive drugs may interfere with certain stages of the memory T-cell response including activation, expansion and trafficking. However, it is possible that other mechanisms by which memory T cell contribute to allograft rejection will remain intact despite immunosuppression resulting in eventual graft loss. It should be noted that combinatorial strategies resulting in the establishment of mixed hematopoietic chimerism were successfully applied in humans and resulted in long-term renal allograft survival and donor-specific tolerance (11,12). While the achievement of mixed chimerism state in transplant patients required a lymphoablative component, the direct effect of these therapies on pre-existing memory T cells has not been analyzed. In the future, careful analysis of memory T-cell survival and functions will be required to rationally apply existing immunosuppressants in sensitized patients.

Alloreactive Memory T Cells and Costimulation

Classical CD28/CD80/CD86 and CD154/CD40 pathways

Acute allograft rejection, limited long-term graft survival and drug-related toxicities remain significant clinical problems despite advances in immunosuppression. Hence, the selective inhibition of donor-specific T-cell responses is an ultimate goal of transplant research. In addition to TCR/MHC interaction, the activation, survival and function of naïve alloreactive T cells require costimulation through CD28/CD80/CD86 and CD154/CD40 pathways. Interference with one or both of these pathways (‘conventional costimulatory blockade’) usually leads to prolonged allograft survival and, in some cases, tolerance in rodent models (13). However, the same approaches applied to nonhuman primates did not achieve the outcome suggested by rodent studies (14). Studies of renal, cardiac, pancreatic islets and skin transplantation in nonhuman primates revealed that CTLA4-Ig and its analog LEA29Y, anti-CD80/CD86 monoclonal antibodies, anti-CD154 antibodies or combinations of the above delayed the onset of allograft rejection but failed to induce long-term graft survival or tolerance. The most prominent results were observed when costimulatory blockade was administered continually and rejection occurred rapidly after treatment was withdrawn.

The potential explanations for these discouraging results include but are not limited to the dramatic differences in the total number of cells that need to be controlled in rodents versus larger species, the variances in kinetics of drug absorption and clearance, and the genetic background (even among mouse strains some are more responsive to therapies than others). In addition, compared to inbred rodents housed in pathogen-free facilities, large animals and humans contain many more alloreactive memory T cells that become activated and express function independently of CD28/CD80/CD86 and CD154/CD40 costimulatory pathways (15). In support of this view, studies in mice have confirmed that donor-reactive effector and/or memory T cells can confer resistance to the effects of conventional costimulatory blockade.

In several models, donor-specific effector and memory T cells were generated through the rejection of a previous allograft. Zhai et al. reported that anti-CD154 monoclonal antibody failed to prolong cardiac allograft survival in animals recently sensitized with skin allografts from the same donors (16). Under these conditions, the resistance to costimulatory blockade was presumably mediated by primed donor-specific CD8 T cells. In another study, adoptively transferred primed donor-specific T cells precipitated cardiac allograft rejection despite treatment with anti-CD154 monoclonal antibody and transfusion of donor spleen cells (DST/anti-CD154), a strategy that efficiently induces long-term heart allograft survival in naïve mice (17). Experiments using purified memory CD4 T cells further demonstrated that this memory T-cell subset could overcome the beneficial effects of DST/anti-CD154 treatment through multiple mechanisms. In addition to being resistant to the tolerogenic effects of the therapy themselves, memory CD4 T cells provided help for the induction of donor-specific CD8 T cells and alloantibody production that contributed to graft destruction (4). These findings indicate that multiple aspects of the donor-specific memory T-cell response must be controlled in order to prevent costimulatory blockade-resistant allograft rejection.

Another series of studies provide proof of principle that T cells originally primed through exposure to microbial agents can cross-react with alloantigens (heterologous immunity) and influence the outcome of a subsequent allograft. Several groups independently demonstrated that infection with LCMV results in the priming of CD8 T cells cross-reactive to alloantigens. These effector or memory CD8 T cells can later prevent long-term allograft survival achieved through various costimulatory blockade-based strategies including stringent approaches leading to mixed chimerism (18). Analogously, infection with Leishmania major primes cross-reactive allospecific CD4 T cells that mediate rapid rejection of subsequent skin allografts despite DST/anti-CD154 treatment (19).

Taken together, these data show that regardless of the priming antigen, both CD4 and CD8 allospecific memory T cells can initiate allograft rejection despite interference with conventional costimulatory pathways. As memory T cells comprise a significant proportion of the alloreactive T-cell repertoire in humans, these findings have direct relevance to clinical transplantation and undermine the efficacy of costimulatory blockade-based treatments in transplant patients.

Alternative costimulatory pathways

In addition to CD28 and CD154, many other molecules with costimulatory function have been identified on T cells. Based on their molecular structure, these receptors can be divided into two groups: the immunoglobulin superfamily, including inducible T-cell costimulator (ICOS) and programmed death-1 (PD-1); and the tumor necrosis factor receptor superfamily, including CD134 (OX40), CD27, CD137 (4–1BB), CD30 and herpes virus entry mediator (HVEM). The role of alternative costimulation in primary immune responses and the generation of a memory T-cell pool is well established. However, it remains unclear whether these molecules are required for the reactivation and functions of pre-existing memory T cells.

ICOS/B7RP-1. ICOS molecule is not constitutively expressed on naïve T cells, but is induced upon T-cell activation and can be detected on the majority of effector and resting memory T cells. Despite high structural homology to CD28, ICOS does not directly interact with CD80 or CD86. The only known ligand for ICOS, B7RP-1, is expressed on antigen presenting cells and has been found in a variety of nonlymphoid tissues such as kidney, heart and liver. Several groups have demonstrated the involvement of this receptor/ligand pair in the regulation of T-cell differentiation and T-cell-dependent antibody isotype switching (20). Studies of autoimmune responses in mice showed that ICOS provides costimulatory signals for primed but not naïve T cells and that blocking ICOS costimulation after the onset of the disease ameliorates symptoms of EAE and collagen-induced arthritis (20).

In transplant models, ICOS deficiency or blockade delayed the rejection of vascularized cardiac allografts (21). The prolonged survival was associated with decreased priming and expansion of donor-reactive T cells as well as inhibited alloantibody production (22). In contrast to conventional costimulatory blockade that has to be administered at the time of allograft placement in order to prevent priming of donor-specific naïve T cells, treatment with anti-ICOS antibody can be started several days post-transplant, after initial T-cell priming has presumably occurred (22). These results are consistent with previous findings in autoimmunity and contact hypersensitivity models, and support the view that ICOS/B7RP-1 interactions are important for the effector stage of the response by previously activated T cells. However, the direct effects of ICOS costimulation on the reactivation and functions of resting memory T cells remain undetermined.

CD134/CD134L. CD134 or OX40 is expressed on activated CD4 T cells. Resting memory CD4 T cells express very low levels of CD134, which is rapidly up-regulated after reactivation. CD134L is also an inducible molecule that is expressed by activated professional APCs and on vascular endothelial cells (23,24). Signaling through CD134 enhances effector CD4 T-cell responses, while blocking this pathway results in decreased cytokine production and help for antibody production. Consistent with these findings, treatment with anti-CD134 or CD134L blocking antibody alleviated immune responses in rodent models of inflammatory disease such as EAE, collagen-induced arthritis and GVHD (23).

Among other functions, one of the consequences of CD134 engagement is up-regulation of anti-apoptotic genes that in turn increases survival of effector T cells and augments the generation of memory T cells (25). Direct evidence of CD134 involvement in memory recall responses comes from a model of allergic lung inflammation, in which interruption of CD134 costimulation at the challenge stage decreased the asthmatic reaction (23). This study further demonstrated that CD134 signaling was not critical for the expansion of memory T cells, but instead promoted their survival after reactivation.

In transplantation, treatment with anti-CD134L antibody synergized with CTLA4-Ig in prolonging rat cardiac allograft survival (26). This combined treatment prevented the expansion of primed alloreactive T cells and inhibited allograft rejection in sensitized recipients. These data suggest that CD134/CD134L interactions may be involved in immune responses by donor-reactive memory T cells, especially when conventional costimulation is blocked or absent.

CD27/CD70. CD27 is expressed on the majority of naïve T cells and is further up-regulated upon TCR stimulation (24). The interaction of CD27 with CD70, expressed on activated T and B lymphocytes, provides costimulatory signals for antigen-driven proliferation of both CD4 and CD8 naïve T cells (27). In rodent transplant models, CD70 blockade abrogates heart allograft rejection by CD8 T cells, but has little effect on CD4 T-cell-mediated rejection (24).

Several groups have previously demonstrated that the generation of functional CD8 memory T cells is severely impaired in the absence of CD27/CD70 interactions (23). The majority of memory T cells express CD27, and recent findings by Yamada et al. point out the importance of this pathway for pre-existing memory recall responses. Notably, in vivo CD70 blockade inhibits expansion of TCR tg allospecific memory CD8 T cells and prevents cardiac allograft rejection mediated by memory CD8 T cells in mice lacking secondary lymphoid organs (28). However, the role of CD27/CD70 costimulation in transplant rejection by sensitized wild-type recipients remains undetermined.

CD137/CD137L. CD137, or 4-1BB, is induced on both CD4 and CD8 T cells after TCR activation and binds to CD137L expressed by activated APCs (23). Costimulation through CD137 during T cell priming does not affect proliferation, but delivers survival signals to the T cells and as a result, enhances the generation of effector and memory T-cell responses (29). Blocking CD137 signaling inhibits symptoms of collagen-induced arthritis after initiation of the disease (30), raising the possibility that this pathway contributes to memory T-cell responses. In transplantation, administration of signaling anti-CD137 antibody accelerates rejection of murine heart, skin and small bowel allografts (24). No direct data are currently available on the relevance of CD137 costimulation to the functions of alloreactive memory T cells in the context of transplantation.

Taken together, these data show that the costimulatory requirements of alloreactive memory T cells are understudied and that targeting alternative costimulatory molecules is likely to improve allograft survival in recipients containing memory T cells. However, several factors have to be considered while interpreting the data on the role of alternative costimulatory pathways in allograft rejection. First, alternative costimulatory molecules are expressed and up-regulated not only on memory T cells but also on newly generated effector cells. Therefore, the beneficial effect of targeting these pathways may not be solely due to inhibiting memory T-cell responses. Second, it is not clear whether memory T cells require any costimulation in addition to TCR signaling for initial reactivation. It is possible that, similar to CD154, alternative costimulatory receptors merely amplify ongoing activation or act as accessory molecules to enhance effector functions of memory T cells, such as cytotoxicity, interaction with endothelial cells and activation of APCs and B cells. Finally, based on the enhanced survival properties of memory T cells, it is highly unlikely that blocking alternative costimulation will result in complete elimination of pre-existing donor-specific immunity. Hence, the remaining memory T cells may initiate graft rejection after blocking agents are withdrawn.

Concluding Remarks

Donor-reactive memory T cells are likely to be present in human transplant patients and contribute to allograft rejection through multiple pathways. Furthermore, memory T cells circumvent the effects of conventional costimulatory blockade on long-term allograft survival and tolerance induction. To date, there is no reliable approach to prevent the deleterious effect of memory T cells in transplantation settings. Promising strategies beyond the scope of this review include targeting growth factors facilitating survival of memory T cells, membrane channels and intracellular signaling molecules preferentially used by memory T cells and manipulating trafficking of memory T cells through chemokine receptor blockade. It is likely that these approaches will have to be used in combination with conventional costimulatory blockade or immunosuppression in order to inhibit responses of newly activated allospecific T cells. Further, understanding of memory T-cell homeostasis, activation requirements and functions will facilitate the design of improved therapies that should prolong allograft survival in sensitized patients.