T Cell Cosignaling Molecules in Transplantation

Abstract

The ultimate outcome of alloreactivity vs. tolerance following transplantation is potently influenced by the constellation of cosignaling molecules expressed by immune cells during priming with alloantigen, and the net sum of costimulatory and coinhibitory signals transmitted via ligation of these molecules. Intense investigation over the last two decades has yielded a detailed understanding of the kinetics, cellular distribution, and intracellular signaling networks of cosignaling molecules such as the CD28, TNF and TIM families of receptors in alloimmunity. More recent work has better defined the cellular and molecular mechanisms by which engagement of cosignaling networks serve to either dampen or augment alloimmunity. These findings will likely aid in the rational development of novel immunomodulatory strategies to prolong graft survival and improve outcomes following transplantation.

Introduction

Transplant rejection is perhaps unique among immunologic processes in that the mammalian immune system has likely not arisen under evolutionary pressure to respond to surgically implanted, MHC-disparate tissue. A key component of the transplant rejection response is the activation and differentiation of alloreactive T cells and subsequent provision of help for donor-specific antibody (DSA), both processes that are carefully controlled by the balance of costimulatory and coinhibitory signaling received during T cell priming. The critically important role of T cell cosignaling pathways in allograft rejection was recently highlighted in an unbiased survey of the transcripts most tightly associated with acute T cell-mediated rejection (Venner et al., 2014). Using an expression microarray approach to interrogate the molecular phenotypes of rejection in 703 renal transplant recipients, Halloran and colleagues identified a prominence of costimulatory molecules (CD28, CD86, SLAMf8, ADAMDEC1) and coinhibitory molecules (CTLA-4, PD-L1) as being prominent pathways upregulated within rejecting allogeneic tissue. These data thus provided an unbiased confirmation that immunologists are indeed “barking up the right tree” in pursuing T cell cosignaling molecules as targets for therapeutic intervention to control transplant rejection. In addition, transplantation is also unique in that unlike the clinical development of autoimmunity or cancer or infection with a viral pathogen, the exact moment at which the immune system is challenged with alloantigen is known. This situation affords the unique opportunity to intervene on alloimmune responses during the priming phrase, a time when the immune response is most sensitive to cosignaling events, and therefore most susceptible to therapeutic manipulation of those events. As a result, the role of cosignaling pathways has been an area of intense focus in the field for over two decades, the recent highlights of which are discussed in the paragraphs below. This large body of work that informs us that while critically important, the multitude of distinct cellular interactions and alternate binding partners that characterize T cell costimulatory pathways render them highly complex. Further detailed understanding of the kinetics, cellular distribution, binding partners, and intracellular signaling networks of cosignaling molecules in alloimmunity will aid in the rational development of immunomodulatory strategies to prolong graft survival and improve outcomes following transplantation.

CD28 Family Members

Targeting CD28 to temper allograft rejection: from the bench to the bedside and back

Seminal work in the early 1990s implicated CD28 as a critical pathway in the elicitation of graft-destructive T cell responses. The pivotal role of CD28 in facilitating alloimmune responses was identified through antibody blockade studies, using either anti-CD80 and anti-CD86 mAbs (Kirk et al., 2001; Lenschow et al., 1995; Pearson et al., 1997) or a CTLA-4 Ig fusion protein (abatacept) (Larsen et al., 1996; Lenschow et al., 1992; Lin et al., 1993; Pearson et al., 1994) to block ligation of CD28. Treatment of animals with these reagents leads to significantly prolonged allograft survival in experimental models of transplantation. Curiously, rejection of MHC mismatched allografts proceeds seemingly unfettered in recipients genetically deficient in CD28 (Yamada et al., 2001). While it remains possible that this effect is explained by compensation of other costimulatory pathways in animals that lack CD28 during development, the impact of loss of CD28 on Foxp3⁺ regulatory T (Treg) cell survival and function is likely a major contributing factor to this finding (Tang et al., 2003; Zhang et al., 2013). The cell-intrinsic role of CD28 on Treg is discussed in depth by Bluestone and colleagues (add citation at production). Still, the ability of CTLA-4 Ig (and its second generation higher affinity variant LEA29Y, or belatacept) to significantly prolong allograft survival in both murine (Larsen et al., 1996) and non-human primate models (Adams et al., 2005; Larsen et al., 2005) led to the relatively rapid translation of into clinical trials, culminating in the FDA approval of belatacept for use in renal transplantation in 2011.

While costimulation blockade-based therapy using belatacept affords significant benefits over current standard of care therapy in terms of minimizing off-target toxicities (Vanrenterghem et al., 2011; Vincenti et al., 2016), an increased incidence and severity of acute rejection episodes was identified in patients being treated with belatacept, both in Phase II and III clinical trials and in early reports as the therapy moved into general use (Vincenti et al., 2010; Vincenti et al., 2005). This rejection appears to be mediated almost entirely by activated T cells, with limited antibody involvement. The mechanisms underlying belatacept-resistant transplant rejection are just beginning to be elucidated. First, one must consider the impact of blocking CTLA-4 on Treg cell (Schmidt et al., 2009). Seminal studies have reported that CTLA-4 functions on Foxp3⁺ Treg cells in a cell-extrinsic manner to inhibit effector T cells by capturing CD80 and CD86 ligands off of the surface of APCs, thereby limiting the strength of CD28-mediated costimulatory signals received by effector T cells during priming (Qureshi et al., 2011). These findings shed new light on two important observations; first, that in the clinical trials of belatacept in transplantation, the higher-intensity regimen was actually less efficacious than the lower intensity regimen (Vincenti et al., 2005), and second, that co-administration of CTLA-4 Ig precipitated graft rejection in a minor histocompatibility mismatch model of graft tolerance that has been shown to be dependent on the action of Treg cells (Charbonnier et al., 2012; Riella et al., 2012). Taken together, these findings provide evidence that blockade of CD80 and CD86 through the use of CTLA-4Ig fusion proteins such as abatacept and belatacept has the unintended consequence of impairing Treg cell function, which can negatively impact allograft survival.

Second, the adult immune system contains a significant fraction of antigen-experienced CD4⁺ and CD8⁺ T cells, and work from several groups over a number of years has suggested that antigen-experienced T cells may be a primary mediator of belatacept-resistant rejection (Trambley et al., 1999; Valujskikh et al., 2002; Weaver et al., 2009). Alloreactive memory T cells may arise following prior transplantation or pregnancy, lymphopenia-induced homeostatic expansion, or pathogen exposure, recognizing alloantigen on the basis of TCR cross-reactivity (Ford and Larsen, 2010; Krummey and Ford, 2012). Memory T cells, particularly within the CD8⁺ compartment, exhibit reduced requirement for CD28 costimulatory signals and in rodent models have been shown to mediate costimulation blockade-resistant rejection (Adams et al., 2003; Kitchens et al., 2012). Further, in humans a subset of memory T cells downregulate the CD28 molecule and become CD28⁻, as a result of increasing age and/or chronic immune activation. CD28⁻ memory T cell populations demonstrate diminished proliferative capacity and increased TCR clonality, yet possess enhanced immediate cytotoxic functions (Azuma et al., 1993; Batliwalla et al., 1996; Weng et al., 2009). Due to their CD28⁻ status these cells would be independent of CD28 costimulation for recall responses and might therefore be predicted to mount anti-donor T cell responses despite therapeutic immunosuppression with belatacept. While this prediction appeared to bear out in in vitro studies of belatacept resistance in allogeneic mixed-lymphocyte reactions (MLR) (Lo et al., 2011), emerging evidence in non-human primate models and in patients treated with belatacept suggests that recipients possessing increased frequencies of CD28⁻ cells are actually less likely to experience acute rejection following treatment with belatacept (M. L. Ford and A. B. Adams, unpublished observations). These results suggest that while CD28⁻ cells may be independent of CD28 costimulation, their terminal differentiation status may render them too senescent to successfully wage a rejection response in vivo. Further investigation aimed at elucidating the functional role of CD28⁻ T cells during allograft rejection in the setting of costimulation blockers is warranted.

Finally, the presence of alloreactive Th17 cells may play a role in mediating belatacept-resistant rejection. While inflammatory CD4⁺ CCR6⁺ Th17 cells are indeed contained within the human alloreactive T cell compartment (Krummey et al., 2013), experimental evidence linking Th17 cells to cellular rejection in the absence of immunosuppression is controversial (Chadha et al., 2011). The conflicting nature of these reports suggests that the role of Th17 cells in allograft rejection may be tissue- and context- dependent. In particular, human Th17 cells exhibit increased resistance to belatacept-mediated inhibition as compared to IFN-γ secreting Th1 cells in vitro and in vivo (Figure 1) (Krummey et al., 2013). This increased resistance was characterized by a dramatic upregulation in the expression of the negative regulator CTLA-4 on the surface of human Th17 cells (Krummey et al., 2013). This finding is important because by binding to CD80 and CD86, belatacept not only blocks positive costimulatory signals through CD28, but also negative coinhibitory signals through CTLA-4 (Larsen et al., 2005). Thus, the markedly increased expression of CTLA-4 on Th17 cells renders them more reliant on this coinhibitory signal for control of the response. In the absence of CTLA-4 coinhibitory signals, Th17 cells may become hyper-activated and mediate graft injury. Two additional lines of evidence exist in the published literature to support this hypothesis: first, studies in mouse models of liver transplantation, where Th17 cell alloimmunity is prominent, have suggested that CTLA-4 coinhibitory signals are in fact required for liver allograft survival (Li et al., 2005), and second, in clinical trials of autoimmune diseases known to be mediated by primarily by Th17 cells including MS and IBD, abatacept (CTLA-4 Ig) was ineffectual or even exacerbated disease (Linsley and Nadler, 2009; Merrill et al., 2010; Sandborn et al., 2012) (the role of costimulation on Th17 cells in autoimmunity is further reviewed by Vignali et al (insert reference at publication). Thus, it is possible that rejection in patients treated with belatacept may function via a break through of Th17 cell immunity.

Figure 1. Mechanisms of belatacept-resistant rejection following transplantation.

Belatacept, a CTLA-4-Ig fusion protein, binds to CD80 and CD86 thereby inhibiting CD28-mediated signaling during the initiation of a donor-reactive T cell response. Mechanisms by which alloreactive T cell responses still occur in the presence of belatacept, precipitating allograft rejection, include: 1) Foxp3⁺ Treg cell function may be diminished due to the inability of CTLA-4 expressed on the Treg cell to ligate CD80 and CD86 in the presence of belatacept; 2) CD28⁻ CD4⁺ or CD8⁺ T cells that arise as a result of increasing age or chronic inflammation may be inherently independent of CD28-mediated costimulation during activation; and 3) Th17 cells express increased levels of CTLA-4 and are highly sensitive to CTLA-4-mediated coinhibitory signals, rendering them more activated when CTLA-4 ligation of CD80 and CD86 is blocked by belatacept.

Selective CD28 blockade: the next generation of costimulation blockers

The limitations of current CD28 costimulation blockers discussed above are all a result of the concomitant blockade of CTLA-4-mediated coinhibitory signals (Figure 1). Thus, the development of reagents to more specifically inhibit CD28, while leaving CTLA-4 intact, would be predicted to better inhibit alloreactive T cell responses. The development of such reagents was stymied by the unfortunate events of the TGN1412 trial, in which an anti-CD28 mAb induced a potentially lethal cytokine storm among participants in a Phase I safety study as a result of CD28 crosslinking-induced TCR-independent activation of memory T cells (Suntharalingam et al., 2006). Advent of the development of Fc-devoid or Fc-silent antibodies over the last decade facilitated the generation of non-crosslinking anti-CD28 blocking reagents (Liu et al., 2014; Poirier et al., 2010; Suchard et al., 2014; Zhang et al., 2011). Use of these “selective” CD28 blockers in murine and non-human primate models of transplantation has indeed revealed superior graft survival results as compared to belatacept, an effect which is dependent on the preservation of negative signaling through CTLA-4 (Liu et al., 2014; Zhang et al., 2011). This preservation of CTLA-4 mediated signals is associated with enhanced Treg cell functionality (Poirier et al., 2010). Furthermore, in one study, preserved CTLA-4 coinhibition in the absence of CD28 costimulation led to the upregulation of the 2B4 coinhibitory molecule on a subset of CD8⁺ donor-reactive T cells, and expression of this secondary wave of coinhibitory signaling was shown to be functionally important for the efficacy of selective CD28 blockade (Liu et al., 2014). It is important to note that while several studies have confirmed that the improved efficacy of selective CD28 blockade in controlling allograft rejection is a result of preserved CTLA-4 coinhibition, altered engagement of other pathways cannot be ruled out. For example, coinhibitory signaling is initiated upon ligation of CD80 to PD-L1 (Butte et al., 2007) (Figure 2), an interaction that would be inhibited in the presence of belatacept but preserved in the presence of selective CD28 blockade. Second, it has been shown that in humans but not mice CD28 can also bind to ICOS-L (Yao et al., 2011) (Figure 2), an interaction that remains intact in the presence of belatacept but would be inhibited by selective CD28 blockade. These interactions between coinhibitory molecules of the CD28 family are further discussed by Sharpe et al (insert reference at publication).

Figure 2. Costimulatory and coinhibitory molecule interactions of the CD28 family.

CD28 family members commonly have more than one ligand, and interactions include cross-talk between costimulatory and coinhibitory receptors. Dotted lines represent interactions between CD28 family members expressed on T cells and APCs, resulting in either costimulatory (+) or coinhibitory (--) signaling into T cells.

Targeting CD28 to prevent development of donor-specific antibody

While numerous studies have demonstrated a pivotal role for CD154-mediated signals in the provision of T cell help to B cells in order to facilitate class switching and affinity maturation, studies using CD28 costimulation blockers abatacept and belatacept have also demonstrated at potent role for CD28 in mediating this function. Experimental models of transplantation in both mouse and non-human primate have revealed that blockade of CD28-mediated signals impairs the generation of donor-specific antibody (Chen et al., 2013; Ford et al., 2007; Kim et al., 2014), which if left unmitigated can result in antibody-mediated rejection and late graft loss. Specifically, in a rhesus renal transplant AMR model, the addition of belatacept significantly attenuated the development of de novo donor-specific antibody, and in situ analysis of lymph nodes from belatacept-treated recipients revealed a reduction of B cell clonal expansion in germinal centers (GC), GC-follicular helper T (T_FH) cells, and IL-21 production inside GCs (Kim et al., 2014). Indeed, clinical trials are underway to explore the utility of using belatacept to prevent sensitization in transplant recipients with failed allografts in order to prevent the generation of anti-HLA antibodies that would impair their ability to receive and accept a second allograft (I. R. Badell, Emory, personal communication).

Moreover, recent studies have also investigated the ability of belatacept to diminish levels of pre-existing donor-specific antibody in pre-sensitized transplant recipients who possess anti-HLA antibodies at the time of transplantation. Knechtle and colleagues have shown that administration of belatacept to NHP who developed donor-specific antibody as a result of prior skin transplantation in the absence of immunosuppression resulted in diminution of donor-specific antibody titers over time, and led to attenuation of antibody-mediated rejection of a subsequent kidney transplant (Burghuber et al., 2015). The biologic basis of this observation is likely a result of the expression of CD28 on plasma cells or short-lived plasmablasts. Seminal studies have shown that even in the presence of sufficient T cell help, loss of CD28 (or ligands CD80 and CD86) results in a profound loss of long-lived plasma cells in the bone marrow, along with a marked inability to maintain long-term antibody titers in the serum (Rozanski et al., 2011). Thus, the impact of belatacept treatment on both pre-existing anti-HLA or anti-donor antibodies, as well as protective antibodies against pathogens, represents an important area of future investigation.

The prospect of selective CD28 blockade likely has important implications for antibody-mediated rejection during transplantation. Several recent studies have highlighted the critical role of CTLA-4 coinhibition in limiting CXCR5⁺ T follicular (Tfh) cell responses, thus impairing the development of high affinity antibodies (Sage et al., 2014; Wang et al., 2015; Wing et al., 2014). Loss of CTLA-4 on effector Tfh cells led to in spontaneous Tfh cell differentiation and exaggerated GC B cell responses in vivo (Sage et al., 2014; Wang et al., 2015). Furthermore, even short-term blockade of CTLA-4 resulted in a significant increase in Tfh cell differentiation and GC development (Wang et al., 2015). Thus, it is possible that next generation selective CD28 blockers that leave CTLA-4 coinhibition intact will more effectively inhibit de novo donor-specific antibody and prevent antibody-mediated rejection during transplantation.

ICOS: A subdominant costimulatory role in transplantation?

Another member of the CD28 family that has been targeted therapeutically transplantation is the inducible T-cell costimulator (ICOS)(Hutloff et al., 1999). Unlike CD28, ICOS is not expressed on resting CD4⁺ or CD8⁺ T cells but is rapidly upregulated upon T-cell activation.(Dong et al., 2001) As discussed above, a complex relationship exists between the ligands for CD28 and ICOS, in that CD28 can also bind to ICOS-L in humans (albeit with lower affinity than ICOS for ICOS-L and CD28 for CD80 or CD86) (Figure 2)(Yao et al., 2010). Early studies in murine transplant models have revealed that ICOS-mediated signals can precipitate both acute and chronic rejection (Nanji et al., 2006; Ozkaynak et al., 2001). For instance, ICOS antagonism synergized with CTLA-4-Ig in attenuating the effector function of donor-reactive memory T cells and prolonging graft survival (Schenk et al., 2009). ICOS also serves as a useful readout of strength of T cell stimulation during transplantation, in that it is significantly downregulated in the presence of CD28 inhibition with CTLA-4 Ig, and even more profoundly in the setting of selective CD28 blockade where CTLA-4 coinhibition is preserved (Liu et al., 2014). Despite the association of reduced ICOS expression with increased graft survival in the presence of selective CD28 blockade, this reduction in ICOS expression was found to not be mechanistically responsible for the diminished rejection response, in that overexpression of ICOS on the surface of donor-reactive T cells failed to enhance alloreactive CD8⁺ T cell responses or precipitate rejection (Liu et al., 2015a). The conclusion that ICOS-mediated signals may play a subdominant role in the initiation and differentiation of alloreactive T cell responses is further solidified by recent evidence that ICOS blockade, either alone or in combination with belatacept, resulted in no measurable effect on allograft rejection in a non-human primate model of kidney transplantation (Lo et al., 2015). However, these findings do not preclude a role for ICOS in the provision of help for the production of alloantibody. Indeed, recent studies have highlighted a pivotal role of donor-reactive memory CD4⁺ T cells in the generation of GC B cell responses independent of the CD154-CD40 pathway (Gorbacheva et al., 2015; Rabant et al., 2013). ICOS plays a known critical role on the recruitment and function of Tfh cells (Xu et al., 2013); in this way the ICOS costimulatory pathway may significantly contribute to the alloimmune response in the context of transplantation.

PD-1: A potent coinhibitory role in transplantation tolerance

In addition to the well defined role for CTLA-4 in attenuating allograft rejection, significant evidence exists to support an important role for PD-1 mediated coinhibition during transplantation, particularly during the maintenance of allograft tolerance. Expression of the transcripts of PD-1, PD-L1, and PD-L2 are all increased upon allogeneic transplantation as compared to syngeneic controls (Ozkaynak et al., 2002), suggesting that engagement of these coinhibitory pathways is a direct result of activation of alloimmunity, and not non-specific tissue injury. Moreover, numerous studies using both pathway blockade and agonism approaches have revealed functional effects of the PD-1 pathway. For example, studies in a fully MHC mismatched model of cardiac transplantation demonstrated that antibody blockade of PD-1 and PD-L1, but not PD-L2, significantly accelerated cardiac graft rejection (Ito et al., 2005). Blockade of PD-1 was also shown to precipitate allograft rejection in a model of surrogate minor antigen disparity (OVA) in which donor-reactive T cells could be tracked over time (Koehn et al., 2008). Treatment with anti-PD-1 mAb resulted in increased donor-reactive cell proliferation and cytokine secretion, even in the presence of CD28 and CD154 blockade. In addition to its impact on acute allograft rejection, PD-1 ligation likely plays an important role in the development and maintenance of allograft tolerance. Early administration of anti-PD-L1 mAb (as well as use of PD-L1 deficient recipients) resulted in abrogation of the establishment of transplantation tolerance in a murine model of cardiac allotransplantation, and was associated with increased IFN-γ secreting effector CD8⁺ T cells as well as a decrease in graft-infiltrating Foxp3+ Treg (Tanaka et al., 2007; Wang et al., 2007). Reciprocally, more recent studies of animals in which transplantation tolerance has been achieved have revealed an association between an increased frequency of PD-1^hi Foxp3⁺ Treg cells and establishment of the tolerant state (Gupta et al., 2012). Thus, the PD-1 pathway is necessary but not sufficient for tolerance induction and sustenance, suggesting that agonism of the PD-1 pathway would need to be combined with some other immunomodulatory strategy to facilitate the development of tolerance in transplantation.

Interestingly, exposure to inflammation at the graft site may play a role in inhibiting the differentiation of PD-1 effector cells during transplantation. A recent report using animals deficient in fucosyltransferase-VII (Fut7^−/−), an enzyme which is required for the biosynthesis of selectin ligands (Lacha et al., 2002), revealed that following transplantation these animals develop exhausted CD4⁺ alloreactive T cell responses, characterized by increased expression of PD-1 (Sarraj et al., 2014). Blockade of PD-1 in Fut7^−/− animals (but not in wild-type controls) rapidly precipitated allograft rejection, suggesting that impaired leukocyte recruitment to sites of inflammation is a novel mechanism leading to PD-1-mediated CD4⁺ T cell exhaustion, and that allograft-associated inflammation may rescue T cells from an exhausted fate.

Further work dissecting individual roles of PD-1: PD-L1 and PD-1: PD-L2 pathways (Figure 2) during transplantation has revealed distinct effects of these two interactions. Interestingly, in a model of MHC class II disparity, blockade of PD-L1 resulted in abrogation of tolerance, while blockade of PD-1 or PD-L2 had no effect (Yang et al., 2008; Yang et al., 2011). Follow up studies revealed that while both PD-1 and PD-L1 blockade functioned to increase alloreactive T cell proliferation and acquisition of effector function, PD-L1 blockade exhibited a more profound effect as compared to PD-1 blockade. In addition, blockade of PD-L1, but not blockade of PD-1, resulted in diminished alloantigen- specific T cell apoptosis. Interestingly, the unique effect observed following PD-L1 blockade was dependent on the presence of Foxp3⁺ Treg cells (Sandner et al., 2005).

These findings were initially attributed to differences in expression of PD-1, PD-L1, and PD-L2, or in the relative half-lives or affinities of the antibodies used in these studies. However, the discovery that PD-L1 has the potential to ligate CD80 in addition to PD-1 (Figure 2) gave rise to a potential alternate explanation underlying this phenomenon; specifically the idea that PD-L1: CD80 interactions might function in a non-redundant manner to control donor-reactive T cell responses (Butte et al., 2007). The distinct role of PD-L1-CD80 interactions during transplantation was further elucidated though the use of a newly developed anti-PD-1 antibody (10F.2H11) that targets only PD-L1-CD80 interactions, leaving PD-L1-PD-1 interactions intact. Selective blockade of PD-L1-CD80 interactions in minor mismatch cardiac allograft model resulted in a marked increase in histological evidence of rejection and a detectable increase in donor-reactive T cell responses (Yang et al., 2011). However, the impact of selective blockade of this receptor: ligand interaction was not as profound as complete inhibition of PD-L1 interactions using the pan-PD-L1 inhibitor (MIH-6). In sum, these data suggest that the functions of the PD-L1-PD-1 and PD-L1-CD80 pathways are at least in part non-overlapping in terms of their abilities to attenuate alloimmunity following transplantation.

PD-1H: New kid on the block

In 2011 a protein possessing strong homology to the IgV region of PD-1 was identified and termed PD-H1 (VISTA)(Flies et al., 2011; Wang et al., 2011) (Figure 2). PD-H1 is widely expressed by hematopoietic cells, but is upregulated on both CD4⁺ and CD8⁺ T cells following antigen encounter. Importantly, therapeutic manipulation of the PD-H1 interactions using the recently described agonistic antibody M5HA resulted in abrogation of graft-versus-host disease in both semi- and fully-allogeneic mouse models of bone marrow transplantation (Flies et al., 2011). Mechanistically, PD-1H coinhibitory receptor signaling exerts its effect by blocking alloreactive T cell proliferation and accumulation following transplantation (Flies et al., 2015). Moreover, PD-H1 agonism results in the expansion and survival of alloreactive Treg cells (Flies et al., 2015). However, the precise cellular interactions involved in this pathway are still being determined: studies in models of tumor immunity revealed that PD-1H expression by CD11c⁺ antigen presenting cells (APCs) can negatively regulate T cell responses via ligation of an unknown receptor (Wang et al., 2011) (Figure 2) but adoptive transfer studies of PD-1H-deficient T cells revealed a T cell intrinsic effect of PD-H1 in inhibiting CD4⁺ T cell responses (Flies et al., 2014). Thus, more work is required to elucidate the mechanism underlying the coinhibitory properties of PD-1H, its role in alloimmunity and T cell tolerance, and the potential for therapeutic targeting of this pathway in transplantation.

TNF Family Members in Transplantation

Unrealized potential: The CD154-CD40 pathway

Next to CD28, CD154-CD40 interactions are perhaps the most well studied pathway in transplantation. A myriad of studies in murine and non-human primate studies over many years have demonstrated a critical role for CD154-CD40 interactions in initiating the alloimmune response, and a profound effect of CD154-CD40 blockade on graft survival (Elgueta et al., 2009; Pinelli and Ford, 2015; Quezada et al., 2004). Inhibition of CD154-CD40 interactions diminishes innate immune responses to transplanted tissue (Ferrer et al., 2012b), including preventing the maturation of alloantigen-presenting dendritic cell (DC) and promotes the generation of a population of tolerance-inducing plasmacytoid DC (Ochando et al., 2006), resulting in diminished expansion and differentiation of allospecific CD4⁺ and CD8⁺ effector T cells. In addition to this potent effect on effector T cell responses, CD154 antagonism promotes conversion of conventional CD4⁺ T cells into Foxp3⁺ peripheral Treg cells (Ferrer et al., 2011) and increases accumulation of Treg cells within the allograft and subsequently in the graft-draining LN (Zhang et al., 2009). Evidence also exists to suggest that antagonism of CD154-CD40 interactions results in the upregulation of coinhibitory molecules on donor-reactive T cell populations, including PD-1 (Rayat and Gill, 2005), KLRG-1 (Ferrer et al., 2012b), and TIM-3 (M. L. Ford, unpublished observations). CD154-CD40 interactions are also critical for the development of donor-specific antibody and antibody-mediated rejection, and therapeutic blockade of this pathway diminishes the development of alloreactive B cell germinal center responses in mice (Chen et al., 2013) and mitigates antibody-mediated rejection of kidney allografts in non-human primates (Kim et al., 2014). Further, blockade of the CD154-CD40 pathway is perhaps the most potent method of long-term tolerance induction yet identified in experimental models of transplantation, in that recipient exposure to resting hematopoietic donor cells (in the form of donor splenocyte infusion or donor bone marrow transplantation) in the presence of CD154-CD40 blocking reagents has been shown to result in durable tolerance to skin, heart, and islet transplantation (Pinelli and Ford, 2015; Quezada et al., 2004). Early studies suggested that the potency of anti-CD154 mAbs in mediating this effect was the result of FcR-dependent clearance of CD154-expressing activated alloreactive T cells (Monk et al., 2003), but more recent reports demonstrated the utility of Fc-devoid anti-CD154 antibodies argue against this mechanism (Daley et al., 2008; Pinelli et al., 2013). Likewise, antibodies directed toward CD40 have demonstrated comparable efficacy to CD154-directed reagents in most models (Badell et al., 2012b; Gilson et al., 2009; Lowe et al., 2012; Thompson et al., 2011b).

While current dogma connotes the expression of CD40 on the APC and CD154 on the CD4⁺ T cell (Figure 3), several studies have examined the potential roles of CD154 and CD40 expressed on other cell types in the context of transplantation. Clearly, CD40 on APC plays an important role in the activation, migration, and cytokine secretion of CD11c⁺ DC that serve to activate alloreactive effector T cells (Ferrer et al., 2012a). However, recent studies have also identified an important cell-intrinsic role for CD40 on donor-reactive CD8⁺ T cells in the context of transplantation (Liu et al., 2013) (Figure 3). Specifically, CD8⁺ T cell accumulation and effector function are blunted and allograft rejection is attenuated when donor-reactive CD8⁺ T cells are genetically deficient in CD40. It is interesting to note that this CD8⁺ T cell-intrinsic role for CD40-mediated costimulation was found not to play a role in the generation of pathogen-elicited CD8⁺ T cell responses (Sun and Bevan, 2004), highlighting a potential difference in the costimulatory requirements during alloimmune responses vs. protective immunity. A potential explanation for this difference is the fact that ligation of toll-like receptors (TLRs) expressed on CD8⁺ T cells in the setting of pathogen infection (Rahman et al., 2008) could compensate for the requirement for T cell intrinsic CD40 signals.

Figure 3. Cellular interactions involving the CD154-CD40 pathway.

Classic T cell licensing involves CD154 expressed on activated CD4⁺ T cells ligating CD40 expressed on DC or other APC, resulting in transmission of an activating signal into the APC (red arrows represent directionality of signaling). CD154-CD40 interactions can also occur through ligation of CD154⁺ CD4⁺ T cells by CD40-expressing CD8⁺ T cells. There is potentially also a role for CD154 expressed on the surface of a CD11c⁺ DC, binding to CD40 on the surface of an activated CD8⁺ T cell. Finally, CD154 expressed on CD4⁺ T cells can also interact with CD11b⁺ monocytes or macrophages.

In addition, CD154 was shown to be expressed on CD11c⁺ DC following TLR ligation (Johnson et al., 2009) (Figure 3). Thus, while this pathway is likely at play during pathogen infection, the role of DC-derived CD154 in the execution of alloreactive immune responses remains to be determined. Lastly, the discovery that CD154 likely has another binding partner—CD11b (Mac-1) (Okwor et al., 2015; Willecke et al., 2014; Wolf et al., 2011; Zirlik et al., 2007) (Figure 3)—sheds significant light on the mechanism by which CD154 blockade potently inhibits aspects of innate cell recruitment during transplantation (Ferrer et al., 2012b). While the contribution of CD154-CD11b interactions to the development of alloimmunity remains an unexplored area, the potential role of this interaction will likely enter into the calculus of whether CD154 or CD40-directed reagents are likely to provide more favorable results for immune modulation in the setting of transplantation.

Given the numerous and potent effects of therapeutic targeting of CD154-CD40 interactions in the setting of alloimmunity, interest in development of pharmacologic inhibitors for translation into clinical transplantation remains high. While early clinical trials were stymied by thromboembolic complications associated with the use of anti-CD154 reagents owing to the expression of CD154 on platelets (Kawai et al., 2000), recent technological developments have resulted in the generation of potentially clinically translatable reagents for targeting this pathway. First, non-agonistic anti-CD40 antibodies are being developed in order to avoid the use of anti-CD154-associated coagulopathic events. While first-generation CD40 blockers functioned at the level of depleting APCs in vivo (Adams et al., 2005; Haanstra et al., 2003; Pearson et al., 2002), newer, non-depleting CD40 blockers have now been generated that show significant efficacy in nonhuman primate models of bone marrow, islet, and renal transplantation (Aoyagi et al., 2009; Badell et al., 2012a; Badell et al., 2012b; Imai et al., 2007; Lowe et al., 2012; Oura et al., 2012; Page et al., 2012; Thompson et al., 2011a). Blockade of the CD154-CD40 pathway has also shown impressive results in prolonging graft survival in non-human primate models of xenotransplantation. Studies show that levels of soluble serum CD154 are significantly elevated following pig to primate xenotransplantation (Ezzelarab et al., 2014), and treatment of NHP recipients of xenogeneic kidneys as well as islet grafts with anti-CD154 monoclonal antibodies significantly prolonged graft acceptance (Higginbotham et al., 2015; Shin et al., 2015). Thus, with a phase IIa clinical trial (NCT01780844) of an anti-CD40 mAb (ASKP1240) currently underway for use in kidney transplantation, therapeutic manipulation of the CD154-CD40 pathway in order to improve outcomes in transplantation may eventually become a clinical reality.

The role of OX40- OX40L interactions

OX40 is another member of the TNF family that is well studied in transplantation, particularly for its critical role in regulating donor-reactive memory T cell responses and regulatory T cells (reviewed in depth by Ware and colleagues, insert reference at publication). OX40 is potently upregulated on both naïve and memory CD4⁺ T cells following transplantation, and blockade of OX40 signals inhibits allograft rejection mediated by these cells (Kinnear et al., 2013; Kinnear et al., 2010). Mechanistically, inhibition of OX40 in vivo was found to have no effect on CD4⁺ T cell proliferation, but rather absence of OX40 signaling diminished alloreactive T cell survival, resulting in overall fewer graft-infiltrating cells in recipients treated with OX40 blockade. Further, studies showed that donor-reactive memory CD4⁺ T cell responses that are relatively insensitive to blockade of either CD28 or CD154-mediated signals during transplantation can be attenuated via OX40 blockade (Vu et al., 2006). On the other hand, agonism of the OX40 pathway can overcome transplantation tolerance induction mediated by CD154 blockade (Burrell et al., 2009). In contrast to its role in augmenting CD4⁺ T effector and memory cell survival, OX40 ligation may inhibit the suppressive capacity of Foxp3⁺ Treg cells. For example, ligation of OX40 on Foxp3⁺ Treg cells results in a loss of the ability of these cells to suppress effector T-cell proliferation and IFN-γ production (Vu et al., 2007) and impairs Treg cell survival (Kinnear et al., 2013). Taken together, these results suggest that therapeutic inhibition of OX40 mediated signals in transplantation may function to simultaneously inhibit CD4⁺ effector response and enhance Foxp3⁺ Treg cell-mediated suppression, suggesting that OX40 may be a promising candidate for targeting in clinical transplantation.

CD2 Family Members

The CD2 family of costimulatory and adhesion molecules has also been shown to play a significant role in the execution of an alloimmune response and thus these molecules have been therapeutically targeted in experimental models of transplantation. CD2 is constitutively expressed all T cells but upregulated upon antigen recognition (Sanders et al., 1988). Importantly, CD2 is more highly expressed on effector memory T-cells relative to central memory T cells (Weaver et al., 2009) and therefore more effectively target those cells that are poised to rapidly exert effector function upon encounter with cognate antigen, cells which are often free from control by other regulatory checkpoints (i.e. CD28). In addition to its role in facilitating adhesion of T cells to APC during the immunological synapse, CD2 ligation (by its integrin binding partner LFA-3 in humans) results in the direct transmission of co-stimulatory signals to promote T cell activation and differentiation (Moingeon et al., 1989; van der Merwe, 1999). As such, CD2-LFA-3 interactions have been therapeutically targeted in translational studies in non-human primate models of transplantation. Borrowing a drug that was FDA-approved to treat plaque psoriasis (Brimhall et al., 2008; Ellis et al., 2001), Weaver et al. showed that alefacept [Au: A brief (one line) description of the drug you mentioned would be helpful.] synergized with belatacept and anti-CD154 mAb (5C8) in prolonging renal allograft survival (Weaver et al., 2009). In vitro analyses of the effect of targeting CD2 on human alloreactive T cells demonstrated that inhibition of CD2-mediated costimulation significantly attenuated the proliferation of CD2^hi belatacept-resistant memory CD8⁺ T cells (Lo et al., 2011). Interestingly, more recent studies of NHP kidney and islet transplantation performed in the absence of CD154 antagonism) showed no additional benefit of the addition of alefacept on graft survival, and instead as associated with a significant increase in the frequency of cytomegalovirus reactivation (Lo et al., 2013; Lowe et al., 2013).

2B4 (CD244, SLAMf4) is also member of the CD2 subset of immunoglobulin superfamily molecules (Lee et al., 2004; Vaidya et al., 2005). Previously best known for its role on NK cells, more recent work has shown that in certain settings 2B4 can be inducibly expressed on CD4⁺ and CD8⁺ T cells (Bengsch et al., 2010; Blackburn et al., 2009; Raziorrouh et al., 2010; Rey et al., 2006; Waggoner et al., 2010; Wang et al., 2010; Wherry et al., 2007). 2B4 possesses both activating and inhibitory functions, which are dependent on the level of 2B4 expression, degree of binding by its ligand CD48, and level of intracellular association with the adaptor molecule SLAM-associated protein (SAP) (Laouar et al., 2007). However, recent evidence in both murine and human cells indicates that its role in T cells is likely coinhibitory (Brown et al., 2011; Kim et al., 2010). We found that genetic deletion of 2B4 solely on donor-reactive CD8⁺ T cells had no impact on the expansion or differentiation of alloreactive T cell responses in the setting of unmodified rejection (Liu et al., 2014). However, in the setting of selective CD28 blockade in which CTLA-4 signals are preserved, 2B4 is upregulated on donor-reactive CD8⁺ T cells, and under these conditions genetic deletion of 2B4 results in augmented accumulation and augmented effector function of these alloreactive T cells, resulting in accelerated rejection of skin grafts (Liu et al., 2014). However, given its potential for multiple binding partners both extracellularly and in terms of intracellular association with various adaptors and phosphatases, more studies are warranted in order to assess the potential for therapeutic manipulation of the 2B4 pathway in transplantation.

The Impact of LFA-1 Antagonism

Ligation of LFA-1 is critical for trafficking of alloreactive T cells into the allograft itself, but can also deliver potent costimulatory signals that augment T cell effector function. For instance, early studies showed that LFA-1 antagonism in the setting of transplantation was effective at inhibiting priming of naive alloreactive T-cell responses and prolonging survival of allogeneic cardiac and islet allografts (Kitchens et al., 2012; Nicolls and Gill, 2006; Setoguchi et al., 2011). Further, LFA-1 antagonism is effective in mitigating allograft rejection mediated by donor-reactive memory T cells that independent of CD28 and CD154 costimulatory signals, in both mouse and non-human primate models (Badell et al., 2010; Poston et al., 2000; Thompson et al., 2012). Interestingly, inhibition of LFA-1 ligation in the presence of mTOR inhibition using rapamycin results in long-term graft survival even following cessation of all immunosuppression. It is interesting to speculate that this observed long-term graft survival is the result enhanced frequency or functionality of Foxp3⁺ Treg cells, as both LFA-1 antagonism and mTOR inhibition have been shown promote Treg expansion and suppressor function in vivo (Reisman et al., 2011; Singh et al., 2012). Based on these findings, LFA-1 antagonism represents a clinically attractive therapeutic target, especially for use patients possessing alloreactive memory T cells, such those receiving a second transplant, or those possessing elevated frequencies of memory T cells as a result of persistent inflammation associated with chronic kidney disease or underlying autoimmunity. In two clinical trials in allogeneic islet transplantation, all patients treated with the anti-LFA-1 monoclonal antibody efalizumab experienced sufficient islet function for the duration of antibody treatment (Posselt et al., 2010; Turgeon et al., 2010; Vincenti et al., 2007). Interestingly, patients treated with efalizumab in combination with the mTOR inhibitor rapamycin exhibited the same striking elevation in peripheral blood Foxp3⁺ Treg cell that was observed in the non-human primate study mentioned above (Posselt et al., 2010). These results support the hypothesis that LFA-1 antagonists might be effectively paired with mTOR inhibitors in order to enhance T-cell regulation and inhibit rejection. Further trials of efalizumab were halted over concerns related to impaired protective immunity when 3 patients developed progressive multifocal leukoencephalopathy—a fatal JC viral disease. Thus, though the future of LFA-1 antagonism for therapeutic manipulation of alloimmunity following transplantation remains to be determined.

Regulation of Alloimmunity by TIM Molecules

TIM-3

Three members of the T cell-immunoglobulin mucin (TIM) family of cosignaling molecules have been explored in the context of transplantation (Yeung et al., 2011). Perhaps the most well-understood of these is TIM-3 (reviewed in depth by Kuchroo, insert reference at publication), which came on the scene in transplantation in 2003 with the seminal study demonstrating that TIM-3 pathway blockade prevented the development of transplantation tolerance induced following CD154 blockade, at least in part via by preventing suppressive function of CD25⁺ Treg cells (Sabatos et al., 2003; Sanchez-Fueyo et al., 2003). More recent work further defined these TIM-3⁺ Treg cells, most of which co-express PD-1, as exhibiting enhanced suppressive function and increased expression of CD25, CD39, CD73, CTLA-4, IL-10, and TGF-β. Administration of the TIM-3 ligand galectin-9 resulted in significantly prolonged graft survival in both murine skin and cardiac transplant models (Cai et al., 2013; He et al., 2009), and was associated with a reduction in both Th1 and Th17 cell effector function and enhanced Treg cell suppressor function (Boenisch et al., 2010). Subsequent investigation in clinical transplantation revealed elevated TIM-3 mRNA levels in rejecting kidney grafts as compared to patients with stable graft function or patients with other causes of allograft dysfunction (Ponciano et al., 2007), raising the possibility that TIM-3 expression could be used as a biomarker to diagnose acute cellular rejection. While this idea may seem to contradict the finding that TIM-3 functions in a coinhibitory manner to limit T cell activation during transplantation, it is interesting to note that although TIM-3 levels were found to be increased in all cases of acute rejection, patients that were unresponsive to treatment possessed lower intra-graft levels of TIM-3, suggesting that strong TIM-3 signaling may help rescue graft function (Renesto et al., 2007).

TIM-3-Gal-9 interactions also play a critical role in dampening the innate immune response that ensues following cold preservation of transplanted tissue (so-called ischemia-reperfusion injury). Early following transplantation of cold-stored livers in mice, TIM-3⁺ CD4⁺ T cells are observed to infiltrate the graft, and treatment with an anti-TIM-3 blocking antibody resulted in increased histopathologic injury and enhanced hepatocellular damage (Liu et al., 2015b; Liu et al., 2015c). In contrast, adoptive transfer of CD4⁺ T cells from TIM-3 transgenic donors or pre-treatment of mice with rGal-9 promoted hepatoprotection against preservation-association liver damage in this model (Liu et al., 2015b; Liu et al., 2015c). Interestingly, anti-TIM-3 mAb did not increase hepatocellular damage in TLR-4 deficient animals, suggesting that TIM-3 may function to dampen inflammation that occurs as a result of TLR ligation during the process of tissue preparation, storage, and surgical implantation (Uchida et al., 2010).

TIM-1

TIM-1 has also been shown to play a role in alloimmunity, governing both alloreactive T and B cell immunity. Initial studies showed that blockade of TIM-1 significantly prolongs survival of heart transplants in a mouse model (Ueno et al., 2008), and agonism of the TIM-1 pathway increased allospecific effector and memory T cells, perturbed Treg cell function (Yuan et al., 2009), and prevented the induction of transplantation tolerance (Degauque et al., 2008). However, TIM-1 is predominantly expressed on B rather than T cells, including the preponderance of IL-10-expressing regulatory B (Breg) cells (Ding et al., 2011). A critical functional role for TIM-1 on Breg cells during transplantation was suggested by the observation that a low-affinity TIM-1-specific antibody known to induce tolerance in WT mice instead accelerated rejection when mice were deficient in B cells (Ding et al., 2011). Furthermore, TIM-1 is also critical for Breg cell induction during transplantation, in that B cells expressing a mutant TIM-1 failed to differentiate into IL-10-secreting Breg cells in the presence of apoptotic cells and failed to attenuate graft rejection (Yeung et al., 2015). Taken together, these findings suggest that TIM-1 plays a direct role in Breg cell induction, and may represent a novel therapeutic target for therapeutic intervention during alloimmunity.

TIM-4

Finally, TIM-4 functions as both a costimulatory molecule and phosphatidlyserine receptor during an alloimmune response. Blockade of TIM-4 expressed on the surface of DC results in increased conversion of conventional T cells to Treg cells, and resulted in prolongation of graft survival (Yeung et al., 2013). TIM-4 likely plays an important role within the transplanted tissue itself, insofar as transplantation of Timd4^−/− cardiac allografts into WT recipients resulted in enhanced survival and increased susceptibility to tolerance-inducing regimens, and further investigation implicated TIM-4 as being important in the regulation of oxidative-stress induced death in tissue-resident macrophages (Thornley et al., 2014). These data suggest that TIM-4 may promote the survival and maturation of APC within a rejecting allograft.

Conclusions

In sum, the ultimate outcome of alloreactivity vs. tolerance following transplantation is potently influenced by the constellation of cosignaling molecules expressed by T cells during priming with alloantigen, and the net sum of costimulatory and coinhibitory signals transmitted via ligation of these molecules. Manipulation of T cell cosignaling pathways may exact a greater effect on alloimmune responses relative to pathogen-elicited responses, owing to the relative dearth of TLR ligands that can also provide ancillary signals to overcome the threshold of positive signaling events required for T cell activation. This situation is fortuitous in that it may provide a “therapeutic window” for manipulation of cosignaling pathways in transplant recipients, one that allows for the blunting of alloreactivity while leaving pathogen-specific protective immune responses intact. Further, T cell cosignaling molecules are an attractive target during transplantation because their tissue distribution is much more limited than that of the targets of current standard-of-care immunosuppression (i.e. calcineurin inhibitors), potentially resulting in significantly reduced off-target toxicities. As such, optimization of costimulation blockade-based regimens for use in transplantation could pave the way for widespread implementation of “elective” transplants, including pancreatic islet transplants for amelioration of type 1 diabetes and vascularized composite tissue allografts (VCA) as an alternative to prosthetic devices. The need is great, and future studies in both basic and translational models will aim to further elucidate the critical costimulatory and coinhibitory signals necessary for tipping the balance from rejection towards tolerance in transplantation.