The Entangled World of Memory T Cells and Implications in Transplantation

Abstract

Memory T cells that are specific for alloantigen can arise from a variety of stimuli, ranging from direct allogeneic sensitization from prior transplantation, blood transfusion or pregnancy, to the elicitation of pathogen-specific T cells that are cross-reactive with alloantigen. Regardless of the mechanism by which they arise, alloreactive memory T cells possess key metabolic, phenotypic, and functional properties that render them distinct from naïve T cells. These properties affect the immune response to transplantation in two important ways: first, they can alter the speed, location, and effector mechanisms with which alloreactive T cells mediate allograft rejection, and second, they can alter T cell susceptibility to immunosuppression. In this review we discuss recent developments in understanding these properties of memory T cells and their implications for transplantation.

Introduction

Approximately 1-10% of T cells in humans possess TCRs which recognize alloantigen¹ and, following transplantation, have the potential to participate in T cell-mediated rejection (TCMR). Of these, approximately 60% are antigen-experienced memory T cells.² Memory T cells differ from naïve T cells in several important ways that render them able to respond faster and better upon re-encounter with alloantigen compared to their naïve counterparts. These changes include increased ability to traffic to sites of inflammation, altered immunometabolism that allows them to proliferate sooner, and epigenetic modifications that result in more potent effector function. However, the fact that memory T cells are poised for rapid response to antigen also affords them reduced reliance on many of the cellular pathways that are targeted by current immunosuppressive regimens, leaving them more resistant to immune modulation. The ways in which these memory T cells are generated in transplant patients, the mechanisms by which they mediate TCMR, and potential strategies to therapeutically target them during transplantation will be discussed below.

Generation of Alloreactive Memory T Cells

Approximately 50% of transplant recipients have been observed to have pre-transplant donor-specific alloreactivity, which is associated with significant increased risk for acute cellular rejection within the first year post-transplantation.^3,4 This alloreactive response is caused by preexisting memory T cells that may have formed from a variety of immune responses. Alloreactive memory T cells may arise via a variety of different mechanisms, and the manner in which they were generated can have potent effects on their subsequent recall potential and effector function upon encounter with alloantigen. Potentially alloreactive memory T cells can form from previous exposure to alloantigen, via TCR cross-reactivity of pathogen-specific T cells, and homeostatic reconstitution and differentiation of naïve alloreactive T cells into memory cells following lymphodepletion induction therapy (Figure 1).

Figure 1. Mechanisms by which alloreactive memory T cells are generated.

Alloreactive memory T cells can arise via a variety of different methods including exposure to alloantigens in the form of blood transfusion, pregnancy, or prior transplants; cross-reactivity of pathogen-specific memory cells; and homeostatic reconstitution of alloreactive memory T cells in a lymphopenic environment.

Alloreactive Memory T Cells from Prior Exposure to Alloantigen

One mechanism by which alloreactive memory T cells can be generated is directly via prior exposure to allogeneic HLA molecules as a result of prior transplantation, blood transfusions, or pregnancy, which can result in humoral and cellular allosensitization.⁵ Pregnancy can function as a temporary unique, tolerized transplantation, as pregnant females usually do not reject fetuses despite HLA mismatch. Pregnancy itself can have an impact on the mother’s immune repertoire by significantly increasing the frequency of CD4⁺ and sometimes CD8⁺ T_CM, T_EM, and activated memory cells.^6,7 Recent studies have found that long-lived fetal-specific CD8⁺ memory T cells develop during pregnancy. However, these cells exhibit selective dysfunction because while they fail to become activated following a second pregnancy, they can contribute to accelerated cellular rejection following skin graft transplantation.⁸

Recognition of Alloantigen by Microbe-Specific Memory T Cells via TCR Cross-Reactivity

Another source of alloreactive memory T cells is through inherent TCR cross-reactivity of pre-existing memory cells that were generated during a prior infection or exposure to environmental antigen. For example, analysis of virus-specific CD8⁺ T cell clones found that ~50% of them were cross-reactive with at least one HLA alloantigen.⁹ A more recent study investigated the allo-responsiveness of EBV- and CMV-specific T cells isolated using 25 different tetramers and measured reactivity to a panel of HLA-typed allostimulators in a mixed lymphocyte reaction based on IFNγ ELISA. CMV-specific CD8⁺ T cells alone generated a strong alloresponse to 6 different HLA molecules. A single HLA allostimulator had the capability of activating memory T cells from multiple viruses, suggesting that HLA-mismatched transplant recipients may already have an extensive polyclonal repertoire of alloreactive memory T cells created in the protective immune responses to viruses.^10,11 Moreover, effector/memory T cells specific for the CMV immediate early protein 1 were associated with acute renal allograft rejection and contained shared TCR β clonotypes based on TCR Vβ sequencing.¹²

In addition to TCR cross-reactivity, microbe-specific memory T cells may augment alloimmunity via linked recognition when the antigens are presented on shared APC. For example, CD8⁺ memory T cell recognition of its cognate antigen, such as a latent viral antigen or autoantigen, on donor cells is enough to prevent the establishment of tolerance without heterologous reactivity to donor MHC. Specifically, ovalbumin (OVA)-specific CD8⁺ memory T cells in C57BL/6 mice that were found to be non-reactive to BALB/c tissue could induce cellular rejection when OVA was expressed on the donor BALB/c tissue, but not when OVA was administered as a perioperative challenge.¹³

Alloreactive Memory T Cells Arise following Homeostatic Reconstitution after Lymphodepletion

Lymphodepletion has become a relatively prevalent form of induction therapy for transplantation, particularly for kidney transplant recipients. However, memory T cells can be resistant to lymphodepletion, potentially due to decreased accessibility of therapeutic depleting agents to tissue-resident memory cells, as well as decreased expression of the targets of some depleting agents on memory T cells. Depletion-resistant T cell populations are comprised of higher frequencies of T_EM cells, which have increased alloreactive potential, but were found to be susceptible to calcineurin inhibition.^14,15 Thus, the impact of lymphodepletion on the memory T cell compartment is two-fold: 1) existing alloreactive memory T cells may be resistant to T cell depletion and therefore preferentially expand following reconstitution, and 2) naïve alloreactive T cells may differentiate into memory T cells in the cytokine-rich environment induced by T cell lymphopenia. A recent comparison of peri- and pre-transplant administration of mouse anti-thymocyte globulin (mATG) found that peri-transplant depletion was ineffective in larger part due to the ability of memory T cell populations being capable of rapid homeostatic replication promoted by post-transplant inflammation than inefficient depletion.¹⁶ This suggests that appropriate timing of lymphodepletion therapy can increase efficacy in depleting pre-existing potentially alloreactive memory T cells. However, surviving memory T cells have a significant replication advantage in the lymphopenic and post-surgical inflammatory environment that could lead to the rapid reconstitution of the T cell population with fewer clones that have an increased likelihood of alloreactivity.^16,17

Memory T Cells in Acute Rejection

T cell-mediated acute rejection is initiated by recognition of alloantigen, via the direct, semi-direct, or indirect pathways. Memory T cell populations may arise that recognize alloantigen through any of these mechanisms. Unlike naïve alloreactive T cells which must be primed in secondary lymphoid organs, memory T cells are less reliant on secondary lymphoid organs for recognition of alloantigen or execution of an allograft rejection.¹⁸ Memory T cells that traffic into allografts and tissue resident T_RM that are activated and proliferate locally¹⁹ may be more likely to be activated via the direct pathway following recognition of donor peptide:MHC complexes on graft parenchymal cells. Central memory T cells that traffic to secondary lymphoid organs where they are activated by recipient-derived APC may thus be more likely to be activated via the indirect or semi-direct pathways.

Mechanisms of Memory T Cell-Mediated Acute Rejection

Memory T cells contain several key attributes that facilitate the execution of acute rejection responses. First, they are more able to access parenchymal tissue, owing to their expression of adhesion molecules and integrins that facilitate extravasation into inflamed tissues.^20-22 In addition to expression of integrins, cognate antigen is necessary for the firm adhesion and transendothelial migration of graft-specific CD8⁺ effector T cells and is independent of G protein- coupled receptor signaling.^23,24 Moreover, memory T cells exhibit enhanced ability to access peripheral tissues that have been exposed to extended cold ischemic storage. In particular, CD4⁺ memory T cell stimulation of donor-derived dendritic cells to produce IL-12p40 homodimers, but not p40/p35 heterodimers, was required for CD8⁺ memory T cell migration and activation following cold ischemic storage. Once activated, the CD8⁺ memory T cells were capable of mediating CTLA4Ig-resistant acute rejection, highlighting the role of DC-CD4⁺ memory T cell interactions in the induction of this complex response to cold ischemia.²⁵

Once T cells have accessed tissue, they perpetuate inflammation in part by secreting inflammatory cytokines. When examining the effects of pre-sensitized CD8⁺ T_EM cells on cardiac allografts, expression of IL-2 and IFNγ was significantly increased and TGFβ was significantly decreased in the transplanted hearts. IFNγ and CXCL9 were also found to be increased in serum. This indicates a significant induction of pro-inflammatory cytokines by sensitized CD8⁺ T_EM cells inducing cellular rejection that was not controlled by cyclosporin A immunosuppression.²⁶ It has also been found in mice that IL-6 deficiency can prolong allograft survival in sensitized recipients, further indicating the contribution of the pro-inflammatory cytokine niche in the induction of cellular rejection by T_EM cells.²⁷ In contrast to naïve T cells, high affinity IL-2 receptor signaling is not required for the activation of T_EM cells. However, high affinity IL-15 receptor signaling is required for CD8⁺ T_CM and T_EM recall responses and an effective target in costimulation blockade-resistant cellular rejection using a blocking antibody for CD122, which is the shared IL-2/IL-15 β chain primarily expressed on memory T cells.²⁸ Recent evidence has shown that the transcription factor Eomes is a key contributor to the cytokine-producing effector function of memory T cells. High expression of Eomes and T-bet in allo-stimulated proliferating CD8⁺ T cells was associated with higher frequency of TNFα⁺ IFNγ⁺ CD8⁺ T cells. Kidney transplant recipients that remained stable contained an increase in Eomes expression that enhances effector function potential, but it was an increased frequency of double positive Eomes^hi T-bet^hi CD8⁺ T cells that correlated with incidences of cellular rejection.²⁹

Memory T cell effector function may also be enhanced by the humoral immune system, specifically by the presence of donor-specific antibodies (DSA). Antibody-induced complement membrane attack complexes trigger IL-1β-mediated activation of endothelial cells, which allows for the trans-presentation of IL-15 to alloreactive T_EM. These membrane attack complex-activated endothelial cells induced the proliferation of CD4⁺ and CD8⁺ memory T cells and increased their polyfunctionality, which results in a more robust cellular response leading to increased cell-mediated rejection pathology.³⁰ In addition to this example in which DSA promotes memory T cell effector function, a critical role also exists for memory CD4⁺ T cell responses in promoting DSA. In particular, memory CD4⁺ T_FH have been shown to accelerate the production of DSA, even under conditions in which the T cell cognate antigen:MHC complexes presented on the B cell does not match the alloantibody specificity, providing unlinked help to activate naïve allo-specific B cells.³¹ These data shown that memory T_FH are able to provide faster, more effective T cell help for the generation of de novo DSA as compared to naïve CD4⁺ T cells that differentiate into T_FH upon primary stimulation, and suggest the interesting possibility that pre-transplant screening for donor-reactive memory T_FH responses could function as an additional screening tool to assess risk of post-transplant development of DSA.

Subsets of Memory T Cells Participating in TCMR

Some memory T cells have been shown to be more associated with acute cellular rejection in transplant recipients. Central and effector memory CD4⁺ and CD8⁺ T cells have been found to be significantly increased within the transplant as well as in lymphoid organs and PBMCs during rejection with CD8⁺ T cells being the dominant infiltrate in the rejecting organ.³²

Belatacept has been limited in its introduction to widespread use due to increased incidence of rejection compared to current standard of care regimens. Several groups have identified specific memory T cell subsets that may be associated with belatacept-resistant rejection. For example, analysis of belatacept-treated patients revealed an increased pre-transplant frequency of CD4⁺ CD57⁺ PD-1⁻ cells in patients who went on to reject vs those that did not.³³ Another study implicated CD28⁺ CD8⁺ CD45RA re-expressing effector memory cells (T_EMRA) as being increased pre-transplant in belatacept-treated patients who went on to reject. Mechanistic studies found that CD28⁺ CD8⁺ T_EMRA rapidly downregulated CD28 upon stimulation, resulting in a predominance of the CD28⁻ CD8⁺ T_EMRA phenotype within graft infiltrates in belatacept-resistant rejection in an NHP model. This suggests that CD28⁺ memory T cells may not be entirely susceptible to belatacept, but these memory cells provide other unique markers that could be targeted in conjunction with the use of belatacept to prevent cellular rejection.³⁴ Subsequent analysis in an independent cohort by another group also found a trend toward increased CD8⁺ T_EMRA in patients who went on to experience belatacept-resistant rejection.³⁵

In addition, resident memory T cells (T_RM) found in peripheral tissues also play a significant role in mediating allograft rejection. Interestingly, it was found that most CD103⁻ and CD103⁺ T_RM were CD8⁺ and donor-derived T_RM were quickly replaced with recipient-derived cells that enabled an aggressive allograft response.³⁶ Recipient-derived T_RM that become established in the graft, whether antigen-specific or polyclonal, are highly functional and have the capability to maintain cellular-mediated rejection within the graft and are persistently exposed to alloantigen, without acquiring phenotypic or functional characteristics of T cell exhaustion.¹⁹ Fu and Sykes review the role of T_RM in solid organ transplantation and rejection further.³⁷ Additional work has also revealed significant roles of tissue-resident innate lymphoid cells that further contribute to rejection within the peripheral tissue, which was recently reviewed by Charmetant et al.³⁸

Of note, repertoire analysis found that the greatest TCR diversity among memory T cells was found in the graft itself, with increased clonality of alloreactive T cells in peripheral blood, providing evidence for a more localized T cell response within the graft. Phenotypic analysis of these graft-infiltrating cells identified them primarily as CD45RO-expressing CD4⁺ and CD8⁺ memory T cells and CD8⁺ T_EMRA.³⁹ One non-traditional population of T cells that could be of interest is CD4^hi CD8^lo T cells. These cells contain effector mechanisms of both traditional CD4⁺ and CD8⁺ T cells with expression of markers like CXCR5 and PD-1 and secretion of IFNγ, IL-4, IL-21, perforin, and granzyme B. These cells have been found to be elevated in peripheral blood during rejection with a marked increase in the frequency of T_EM within this population.⁴⁰

Due to this significant role in acute rejection, memory cells have become targets that could be used as a predictor for risk of cellular rejection. One rapid flow cytometric whole blood assay was designed to determine potential alloreactivity among 1,491 stimulatory reactions. This assay found that most potentially alloreactive T cells from transplant candidates were CD8⁺ T_EM and susceptible to steroid and calcineurin inhibition. One participant demonstrated some alloreactivity to her donor using this blood assay and was the only tracked participant to develop cellular rejection within the first year after transplant, indicating that detection of potentially alloreactive cells in peripheral blood can be a potentially accurate predictor of cellular rejection potential.⁴¹

Memory T cells in Lymphodepletion Induction Therapy and Reconstitution

T cell depletion using thymoglobulin is a mainstay of induction therapy used as standard of care for many renal, heart, and lung transplant recipients. Importantly, many studies have shown that memory T cells exhibit increased resistance to T cell depletion compared to their naïve counterparts. For example, a recent NHP study found that naïve T cells were more susceptible to rATG depletion as compared to bulk memory T cells, and that T_EM cells were the most frequent memory T cell subset upon reconstitution. In a recent human study, the most prevalent subset upon reconstitution was reported to be T_CM cells.⁴² In a recent report examining the effect of ATG on memory T cell populations in pediatric kidney transplant recipients, naïve CD4⁺ and CD8⁺ T cell populations decreased with an increase in CD4⁺ T_EM and CD8⁺ T_CM compared to non-ATG-induced patients. ATG groups were also found to exhibit a significant increase in the frequency of terminally differentiated subsets as CD57^+/−PD-1⁺ in both CD4⁺ and CD8⁺ populations. Interestingly, patients who received ATG had higher frequencies of CD4⁺ Tregs 1-3 months post-transplantation, but frequency decreased to below the level observed in non-ATG patients after 6 months.⁴³ ATG has also been shown to promote T cell exhaustion that is beneficial to establishing a tolerogenic state, findings that were recently reviewed by Angeletti et al.⁴⁴ These data clearly show that memory T cells are more resistant to T cell depletional therapy, but the mechanisms underlying this are not well-elucidated. Recent studies have found that target antigen downregulation may be one strategy memory T cells use to avoid antibody-mediated depletion, as was reported in the case of anti-CD52 and ATG depletional therapy.⁴⁵ CD4⁺ and CD8⁺ naïve T cells were enriched in anti-CD2-treated groups, including the CD4⁺ Treg population, while groups treated with rATG or anti-CD52 experienced a significant depletion of their naïve populations but memory T cells to a lesser extent.^45,46 The propensity of memory T cells to reside in peripheral tissues (as in the case of T_RM) or in bone marrow niches that are relatively inaccessible to the circulation may also effectively shield these cells from the effects of depleting antibody.⁴⁷

In addition to their relative resistance to antibody-mediated deletion, memory T cells also exhibit a survival advantage during homeostatic reconstitution. This is evidenced by a recent report that found that adoptive transfer of naïve vs. memory T cells into lymphopenic mice resulted in a competitive advantage of the memory T cells, as their co-utilization of IL-7 and IL-15 increased their ability to compete against naïve T cells, which use IL-7 alone. Despite higher expression of IL-7Rα, memory T cells had less potent activation of STAT5 downstream, which gave naïve T cells a replication advantage early in reconstitution that was quickly outpaced as the memory cells used IL-15 from tissues.¹⁷ Moreover, homeostatic proliferation of memory T cells following depletion using mATG has recently been shown to significantly rely on IL-1β and IL-6 from B cells, decreasing the efficacy of T cell depletion in situations in which B cell populations are spared.⁴³ Upregulation of macrophage-inducible C-type lectin (Mincle) in B cells induced this production of pro-inflammatory cytokines. Increased memory T cell homeostatic repopulation promoted by CD40 agonism was found to be abrogated in Mincle-deficient B cells, indicating that induction of memory T cell proliferation by B cells after CD40 ligation occurs through a Mincle-dependent pathway.⁴⁸ Interestingly, B cell depletion using an anti-CD20 mAb prevented CD4⁺ memory T cell alloresponses, but enhanced CD8⁺ memory T cell alloresponses in a mouse skin graft model. This B cell depletion also prevented memory T cell alloresponses after bone marrow transplant, indicating that B cells promote the establishment of chimerism and alloreactive T cell deletion in hematopoietic stem cell transplant.⁴⁹

Targeting Memory T cells in Prolonging Graft Survival

T cells are the primary targets for many available immunosuppressive drugs. However, the different mechanisms by which these reagents inhibit alloreactive responses can result in significant variation in effectiveness against different T cell subsets. Some of these mechanisms include targeting TCR signaling, costimulation/coinhibitory molecules, cell adhesion molecules, and cytokine production (Figure 2).

Figure 2. Therapeutic targets on memory T cells.

Unique aspects of memory T cells that can promote their survival, function, and trafficking including surface marker expression, metabolic profiles, and epigenetic modifications provide potential therapeutic targets for transplantation.

Use of Steroids and mTOR Inhibitors

Glucocorticoids are commonly administered to reduce inflammatory mediators of cellular rejection. However, it was found that high dose glucocorticoids administered to lung transplant recipients experiencing acute cellular rejection was effective at reducing the expression of cytotoxic mediators in donor-derived CD8⁺ allograft-resident memory T cells. However, treatment with steroids did not affect the persistence of these cells and they were still able to accumulate in the graft.⁵⁰

Administration of rapamycin has been shown by many investigators to promote the expansion of CD4⁺ Tregs; but has been shown to be unable to inhibit the replication of memory T cells, even when combined with donor-specific transfusion (DST) in mice.⁵¹ However, mammalian target of rapamycin (mTOR) inhibition via everolimus, in the context of costimulation blockade resistant rejection that was non-responsive to corticosteroids and ATG, was able to effectively inhibit the replication and effector function of CD8⁺ T_EM cells.⁵² These data suggest that mTOR inhibition as the sole immunosuppressive is not effective at inhibiting alloreactive memory responses, but it could be effective in combination with costimulation blockade as a rescue treatment for acute cellular rejection or even as maintenance immunosuppression.

Targeting T cell Costimulatory and Coinhibitory Molecules

The most common costimulation blockade target is the CD28 pathway with three widely studied pharmacologic inhibitors that differ in their blocking strength of the interactions between CD80/86 and CD28/CTLA4: belatacept, abatacept (CTLA4-Ig), and anti-CD28dAb. The widespread use of costimulation blockade has been stinted due to increased incidences of acute cellular rejection and frequency of graft loss compared to use of standard immunosuppression with tacrolimus in kidney transplant recipients.^35,53 This increased risk of rejection is typically associated with impaired ability to inhibit T_EM cells.⁵³ However, one study found that the increased rejection risk in belatacept patients did not correlate with changes in the CD4⁺ or CD8⁺ terminally differentiated T_EM cells ³⁵, suggesting that CD28 pathway blockade may be reasonably effective at controlling these populations. In fact, delayed CTLA4-Ig treatment in mice was able to induce long-term graft survival after T cell-mediated cellular rejection despite an inability to prevent the accumulation of alloreactive memory T cells in the spleen. This suggests an important role of CD28 blockade in the effector function, but not proliferation, of CD8⁺ memory T cells, as cytolytic activity and IFNγ production were significantly impaired ⁵⁴. Interestingly, both CTLA4-Ig and anti-CD28dAb significantly decreased proliferation of alloreactive T cells in the graft-draining lymph nodes, but CTLA4-Ig was less effective at impairing cytotoxic effector functions than anti-CD28dAb.⁵⁵ This is likely due to the coinhibitory role of CTLA4 expressed on memory T cells; insofar as preservation of CTLA4-mediated coinhibitory signals via the use of selective CD28 blockade may better inhibit memory T cell function as compared to CTLA4-Ig which blocks both CD28 and CTLA4.

In addition, emerging evidence from rodent models testing the impact of selective CD28 blockade in the context of T cell depletion suggest that blocking CD28 signals during homeostatic reconstitution mitigates the lymphopenia-induced generation of T_EM and T_RM post-transplant, as well as promoting a memory phenotype CD8⁺ Foxp3⁺ T cells (Ford and Habib, unpublished observations). Additional work to uncover the role of CD28 signaling on memory T cells during homeostatic reconstitution is warranted.

While CD28 blockers are the only costimulation-based immunosuppression FDA approved for use in transplantation, other costimulation molecules have also shown promise as targets in inducing long-term graft survival in preclinical models. Another widely studied pathway for immunosuppression is the CD154-CD40 pathway. Studies from many groups across decades have shown that anti-CD154 can induce long term graft survival.^56,57 However, it was reported that memory T cells in sensitized recipients are resistant to CD154 costimulation blockade, limiting its potential effectiveness as a maintenance monotherapy.⁵⁸ However, the use of lymphodepletion with CD154-CD40 pathway blockade could be a way to eliminate this pool of costimulation blockade resistant memory T cells. Importantly, it has been shown that CD4⁺ T cell help via CD40 signaling to B cells to produce cytokines is required for the rapid homeostatic replication of lymphodepletion-resistant CD8⁺ memory T cells.^43,48,59 Therefore, lymphodepletion in conjunction with CD154 pathway blockade can be highly effective immunosuppression and CD154 blockade has been shown to improve lymphodepletion outcomes by promoting reconstitution via thymic output. Anti-CD154 has also been found to impair the establishment of alloreactive memory T cells as determined by the significant decrease in KLRG1^loCD127⁺ memory precursor T cells.⁶⁰

However, clinical application of early anti-CD154 mAb reagents was halted due to risks of thromboembolism.⁶¹ Attempts to avoid the crosslinking effects of anti-CD154 by targeting its receptor, CD40, elicited promising levels of survival in non-human primates with similar effects on memory T cells as anti-CD154.^62,63 Of note, direct comparison of anti-CD40 monoclonals with anti-CD154 monoclonals revealed reduced efficacy in the anti-CD40-treated animals in terms of long term graft survival.⁶⁴ Mechanistic studies in rodent models revealed that the relative inferiority of anti-CD40s compared to anti-CD154 is attributable at least in part to the ability of CD11b to serve as an alternate receptor for CD154.⁶⁴ CD154:CD11b interactions resulted in increased accumulation of both adaptive and innate immune cells in allograft tissue.

Other costimulatory pathways have increasingly been investigated as potential targets for immunotherapy. These include OX40-OX40L and CD27-CD70. Anti-OX40L increased graft survival on its own in a sensitized mouse model and was highly effective in conjunction with DST or rapamycin and low-dose IL-2.^65,66 These studies found that OX40 pathway blockade significantly decreased memory T cell survival and promoted the generation of CD4⁺ and CD8⁺ Tregs. Anti-CD70 was used to inhibit the CD27-CD70 pathway and significantly increased corneal allograft survival in mice. Anti-CD70 treatment resulted in the inhibition of IFNγ and significant decrease of CD4⁺ memory T cells present in the draining lymph nodes.⁶⁷

Coinhibitory molecules have also been increasingly investigated as potential therapeutic targets. Expression of 2B4, a CD2 family coinhibitory molecule that binds CD48, on memory T cells was associated with lower incidence of rejection in kidney transplant recipients.³³ In mice, it was determined that 2B4 expression limits T cell glycolytic capacity and consequently the proliferation of CD8⁺ memory T cells.⁶⁸ Another coinhibitory molecule, TIGIT, was found to be expressed on CD4⁺ and CD8⁺ memory T cell subsets at-risk of being resistant to belatacept, implicating it as a potential target in costimulation blockade resistant rejection. In vitro culture with agonistic anti-TIGIT antibodies was found to significantly increase memory T cell apoptosis via an increase in FOXP3⁺ Treg suppression activity.⁶⁹ These recent papers highlight potential therapeutic targets and the benefits of potentially pairing costimulation blockade with agonists of coinhibitory pathways to better regulate the memory T cell alloresponses during transplantation.

Targeting Other Pathways Involved in Memory T Cell-Mediated Rejection

Costimulatory molecules are not the only promising targets for immunosuppression of memory T cells. Other potential targets could include those relevant to the stimulation of T cells, including adhesion molecules, other signaling receptors, and cytokines. CD45 is a phosphatase necessary at the APC:T cell immunological synapse for TCR signaling and has multiple isoforms. Canonically, naïve T cells are known to express the high molecular weight isoforms CD45RA, B, and C and memory T cells can be differentiated by the expression of the low molecular weight isoform CD45RO with the assumption that CD45RA, B, and C are fully down-regulated upon CD45RO expression.^70,71 Interestingly, it was found that activated CD8⁺ T cells contained distinct CD45RB^hi and CD45RB^lo populations following an acute viral infection. High CD45RB expression on memory T cells was associated with long-lived high affinity antigen-specific cells following viral infection, and increased potential to mediate an alloreactive T cell response.⁷² Treatment with anti-CD45RB and DST prevented memory T cell-mediated rejection in pre-sensitized mice, suggesting it could be a potential target to inhibit early acute rejection.⁷³ Sphingosine-1-phosphate receptor 1 (S1PR1) signaling is a key mediator of T cell trafficking, which can be blocked by the S1PR1 antagonist FTY720. Of note, FTY720 combined with CTLA4-Ig or tacrolimus significantly increased graft survival, primarily by decreasing the frequency of T_CM and T_EM cells in the periphery and delaying overall T cell recruitment into the graft in both pre-sensitized mouse models and NHPs, which suggests it could function to regulate the alloreactive memory T cell response even when administered only as a short-course therapy following transplantation.^74,75

FcγRIIB is an inhibitory Fc receptor expressed on a subset of CD8⁺ memory T cells ^76,77. It was observed that an increase in FcγRIIB⁺ CD8⁺ memory T cells was associated with decreased risk of allograft rejection and its deletion on alloreactive memory cells led to increased production of proinflammatory cytokines and increased rejection ^76,77. This highlights a non-traditional inhibitory receptor with therapeutic potential to impair memory T cell-mediated rejection. T cell-mediated rejection induces a highly proinflammatory alloimmune response, including the production of proinflammatory cytokines. One study found that ATG-resistant rejection in intestinal transplant recipients was significantly mediated by TNFα⁺ IL-17⁺ memory helper T cells. Targeting this resistant population using anti-TNFα (infliximab) was successful in resolving all patient cases of ATG refractory rejection in the study, suggesting that targeting proinflammatory cytokines could be extremely beneficial for resolving alloimmune responses.⁷⁸ Many of these new therapeutic targets show significant promise, particularly in their potential to function as secondary immunosuppression with costimulation blockade. This offers multiple avenues worth exploring that can specifically target costimulation blockade-resistant alloreactive memory T cells.

Immunometabolism and Inhibition of Anti-Apoptotic Pathways

The intersection of T cell metabolism and cell differentiation is increasingly well-recognized. Upon antigen recognition, the massive and rapid expansion of naïve cell populations is fueled by a switch from oxidative respiration (OXPHOS) to glycolysis. Seminal studies have shown that the switch from effector to back to resting memory state is commensurate with decreased glycolytic function and increased OXPHOS ⁷⁹ (Figure 2). Moreover, the increased mitochondrial mass and capacity for OXPHOS of memory as compared to naïve T cells may underlie the ability of secondary effectors to proliferate and mediate effector function more rapidly than primary effectors⁸⁰. This increased mitochondrial mass was dependent on IL-15, which supported mitochondrial biogenesis and increased expression of a key enzyme involved in fatty acid oxidation⁸¹. Overall, these studies illuminated the fact that increased mitochondrial mass in memory vs. naive T cells confers a bioenergetic advantage that underlies the potential of memory T cells for rapid recall responses. Overall, by revealing their increased reliance of memory T cells on oxidative phosphorylation, these reports demonstrated that T cell metabolism may be a potential therapeutic target for better inhibiting donor-reactive memory T cell responses in the context of transplantation.

Memory T cells also exhibit increased expression of the antiapoptotic protein Bcl-2 (Figure 2). This may render them less reliant on T cell survival signals such as costimulation and common gamma chain cytokines. As a result, targeting memory T cells using inhibitors of Bcl-2 may be an effective therapeutic strategy to control unwanted donor-reactive memory T cell responses in transplantation. ABT-737, an inhibitor of Bcl-2 and Bcl-xL, has been found to induce apoptosis in alloreactive CD8⁺ T_CM and T_EM cells and suppress the allogeneic memory response. Interestingly, the alloreactive memory T cell deletion induced by Bcl-2 inhibition was adequate to restore sensitivity to costimulation blockade and prevent costimulation blockade-resistant rejection in a memory model.⁸² Also, donor-specific Tregs were found to be resistant to the apoptotic effects of ABT-737 similar to observations in naïve T cells, further promoting a tolerogenic state.⁸³ This work highlights the potential of Bcl-2 inhibition to specifically target alloreactive memory T cell populations without significantly impacting regulatory T cell responses.

Impact of Immunosuppression on Memory in Protective Immunity

One major concern in transplantation is the increased risk of infections due to chronic immunosuppression. This risk arises from both the inhibition of naïve immune responses as well as the impairment, or sometimes deletion, of memory cells by immunosuppressive drugs, a situation which has resulted in the exploration of the effects of transplantation and immunosuppression on the anti-microbial T cell landscape and its functionality during infection, as well as developing unique ways to improve protective immunity in transplant recipients.

First, it is well-established that current tacrolimus-based immunosuppressive regiments result in increased risk of infection in transplant recipients. Preservation of memory T cell responses are critical to preserving protective immunity in these individuals. For example, a recent study reported that upon completion of standard antiviral prophylaxis, the CMV-specific CD8⁺ T_EM cell population continually increased during a year post-transplant without detectable viremia, indicating antigenic exposure without active disease continues to influence the CMV-specific T cell population post-transplant and induction of immunosuppression.⁸⁴ Another recent report found that seronegative patients possessed CD137⁺IFNγ⁺ CD4⁺ and CD8⁺ CMV-specific memory T cells, which were found to be protective against CMV viremia, particularly when they received a transplant from a seropositive donor, further highlighting the importance of memory T cells in the protection against such a common problematic viral infection in immunosuppressed transplant recipients.⁸⁵ Additionally, CMV presence can modify the overall T cell landscape in seropositive individuals, which includes an overall decrease in CD4/CD8 T cell ratios, increased expansion of T_EM and T_EMRA populations, increased expression of NK cell markers LIR-1 and KLRG1 on CD4⁺ and CD8⁺, and increased IFNγ and cytotoxic responses.⁸⁶ This potentially contributes to premature immune aging correlated with CMV seropositivity within 1 year post-transplant that was characterized by shorter telomeres and an increase in frequency of terminally differentiated CD57⁺ CD8⁺ T cells.⁸⁷

Viral-specific memory T cell responses are differentially impacted by different classes of immunosuppressants used in transplantation. One study found that belatacept selectively inhibited allo-immune responses without inhibiting virus-specific memory T cell responses against EBV, whereas calcineurin inhibitors fully prevented IFNγ production in virus-specific memory T cells in kidney recipient and healthy donor PBMCs in vitro. This confirmed that calcineurin inhibition was capable of impairing all T cell responses, whereas belatacept blocks naïve T cell activation that requires CD28 costimulation and does not interfere with CD28-independent memory T cell responses.⁸⁸ Likewise, sirolimus mTOR inhibition may also impair naïve T cell response, but it did not impair the CMV-specific CD8⁺ memory T cell response. CD8⁺ memory cells exposed to sirolimus were fine-tuned via IL-2 and STAT5 signaling and expressed increased levels of IFNγ and granzyme B compared to untreated controls.⁸⁹

The effect of immunosuppression on responses to COVID-19 infection and vaccination have been of particular interest in recent years. No significant differences in neutralizing antibody levels and T cell responses between immunocompetent controls and immunosuppressed transplant recipients were found up to one year post-infection. It was also noted that the majority of the CD4⁺ and CD8⁺ T cell responses were specific to the spike, membrane, and nucleocapsid proteins of SARS-CoV-2.^90,91 Of note, one study noticed that the transplant patients demonstrated signs of premature immune aging, similar to what has been observed to be more profound in CMV seropositive transplant recipients.⁹¹

Effect of Protective Immunity on Alloreactivity and Graft Rejection

While immunosuppression can potently impact anti-microbial memory T cell responses, the elicitation of an anti-microbial protective immune response can likewise impact an alloimmune response and graft rejection. Seminal studies in the early 2000s showed that prior infection with viral pathogens resulted in the elicitation of costimulation blockade-resistant allograft rejection.⁹² In contrast, other studies found that maturation of the T cell repertoire induced by clinically significant viral infections including murine polyomavirus, CMV and gammaherpesvirus 68 was not sufficient to induce costimulation blockade resistant cellular rejection. Despite observing the expected increase in T_EM cells, which typically do not require the costimulation signals that CTLA4-Ig inhibits, these infections did not induce heterologous alloreactivity or rejection in heterotopic heart allografts.⁹³ These results can be reconciled by reports demonstrating that qualitative aspects of the microbe involved in eliciting the cross-reactive memory T cell response influence the relative resistance of those memory T cells to immunosuppression. For example, memory Th1 clones were found to be susceptible to costimulation blockade-based immunosuppression, while Th17 clones of the same specificity were resistant.⁹⁴ Finally, the affinity with which a microbe-specific TCR recognizes the allo cross-reactive epitope also influences the relative barrier posed by a given memory T cell population.⁹⁵

Strategies to Improve Anti-Microbial T Cell Responses in Transplant Patients

Prior to the start of immunosuppression, vaccination of transplant candidates against problematic infections can be undertaken in order to minimize infectious risk post-transplant. However, the immunologic effects of the patient’s underlying condition, such as end stage organ disease, has the potential to significantly impair any ability to effectively mount an immune response and generate functional memory following vaccination. For example, because patients with end-stage renal disease exhibit impaired immune responses to peptide vaccines, one study sought to determine the efficacy of a live-attenuated vaccine for varicella zoster virus (VZV).^96-98 Results showed that 77% of end-stage renal disease patients developed a detectable antibody response and 82% developed antigen-specific CD4⁺ T_EM and T_CM cells, which persisted more than a year post-transplantation, suggesting that this could be an effective preventative measure against problematic infections in transplant recipients when such vaccines are available.⁹⁸ EBV-derived virus-like particles were found to induce an IFNγ-mediated memory T cell response in vitro using kidney recipient and donor PBMCs, suggesting it could be a potential form of immunization for immunocompromised transplant recipients to prevent infection or reactivation of EBV.⁸⁸ The impact of vaccination on the development of allo-crossreactive memory T cell responses, however, is largely unknown.

Development of memory T cell-targeted therapies to improve protective immunity

Another strategy to improve protective immunity while limiting alloreactivity has been to keep patients on standard of care immunosuppression while providing exogenously prepared anti-viral T cells to provide immunosurveillance and combat transplant-associated infections. For example, adoptive T cell therapy created by expanding T cells stimulated with 30 HLA class I and II-restricted CMV epitopes has been effective in treating drug-resistant or recurrent CMV in transplant recipients. Responders were observed to exhibit a dramatic shift in their TCR Vβ repertoire, which consisted of reconstitution of both transferred and endogenous clonotypes that were primarily cytokine-producing T_EM with some responders generating a mixture of both T_EM and T_CM phenotypes.⁹⁹ Another group reported the use of a rapamycin-treated autologous T cell product and used CMV as a model virus to determine the effectiveness. They found significantly increased frequencies of highly functional CD4⁺ and CD8⁺ T_CM cells in treated individuals, irrespective of whether the T cell product was generated before or after transplantation and exposure to immunosuppression.¹⁰⁰ Similarly, monocyte-derived dendritic cells cultured with common γ chain cytokines including IL-7, IL-15, and IL-21 have also been used to prime and expand antigen-specific CD8⁺ T_EM cells against CMV in vitro. The addition of these cytokines to PBMC-based stimulations significantly promoted the replication of both naïve-like and CD8⁺ T_CM cells, highlighting the use of cytokines to help fine-tune the generation of anti-viral T cell products.¹⁰¹

Novel immunosuppressive pathways that impair alloreactive T cells but not anti-microbial T cells

It is clear from the study of memory T cells across models of immunity that there is a great deal of heterogeneity within the memory T cell compartment of a given individual, and moreover that populations of memory T cells may thus be more or less reliant on a given molecular pathway for survival, proliferation and effector function. This observation has given rise to the idea that memory T cells responding to an allogeneic stimulus may have distinct requirements for reactivation as opposed to T cells stimulated by an infectious stimulus. For example, alloreactive CD8⁺ T cells require CD154-CD40 for activation, whereas this is dispensable for the activation of Listeria-specific CD8⁺ memory T cell responses.^102,103 Likewise, mTOR inhibitors such as sirolimus inhibit the proliferation of T cells responding to an allogeneic stimulus, while memory T cell differentiation in response to an infectious stimulus or vaccine is enhanced upon exposure to sirolimus.^104-106 A recent report also demonstrated a critical role for Coronin-1 in the elicitation of alloreactive effector T cell responses, while protective immune responses were intact in the absence of Coronin-1.^107-109 Finally, alloreactive effector/memory T cells may be more reliant on IL-7 signaling in that they exhibit less downregulation of the IL-7 receptor CD127 following encounter with antigen than do microbe-stimulated T cells.¹¹⁰ These examples highlight the fact that there may be qualitative differences in the activation pathways between alloantigen-elicited and microbe-elicited T cell responses. Immunotherapeutic approaches aimed at targeting such pathways may better preserve protective immunity while inhibiting unwanted alloimmunity.

Conclusion

The studies discussed in this review highlight the high degree of heterogeneity among memory T cells, in terms of their differentiation, execution of effector function, and expression of potential targets for immune modulation. This heterogeneity invokes a critical need for better assays that could be implemented clinically to identify not only the quantity of donor-reactive memory T cells, but also the quality of the donor-reactive memory T cell compartment in a given patient prior to transplantation. Identification of both quantitative and qualitative aspects of donor-reactive memory T cell populations will allow for better stratification of risk of memory T cell-mediated rejection in a manner similar to the way in which clinical measurement of pre-transplant DSA is currently used to assess risk of humoral sensitization to a given prospective donor. Assessing donor sensitization within the cellular immune compartment in addition to the humoral immune response would facilitate more personalized immunosuppression and has the potential to improve outcomes in transplant recipients.

Abbreviations

APC

Antigen presenting cell

ATG

Anti-thymocyte globulin

CMV

Cytomegalovirus

DC

Dendritic cell

DSA

Donor-specific antibodies

DST

Donor-specific transfusion

EBV

Epstein-Barr virus

ELISA

Enzyme-linked Immunosorbent Assay

HLA

Human leukocyte antigen

MHC

Major histocompatibility complex

mTOR

mammalian target of rapamycin

NHP

Nonhuman primates

OVA

Ovalbumin

PBMC

Peripheral blood mononuclear cells

T_CM

Central memory T cell

TCMR

T cell-mediated rejection

TCR

T cell receptor

T_EM

Effector memory T cell

T_EMRA

CD45RA re-expressing effector memory T cells

T_FH

Follicular helper T cell

Th1

Type 1 helper T cell

Th17

IL-17-producing helper T cell

Treg

Regulatory T cell

T_RM

Resident memory T cell

VZV

Varicella zoster virus