Immune Tolerance for Autoimmune Disease and Cell Transplantation

2.1.1. Ex vivo antigen coupling to cells

The peptide association with erythrocytes builds on decades of research on the coupling of peptides to donor splenocytes (SPs) using the chemical cross-linker 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (ECDI) (34). ECDI is employed to affix antigens to the donor cells in order to minimize the amount of free antigen in the blood that may lead to anaphylaxis; it also induces apoptosis of the donor splenic leukocytes. Antigen coupled to splenocytes (Ag-SP) delivers the antigen to APCs that can mediate immune tolerance. The nonspecific cross-linking of antigen to the cell surface during the induction of apoptosis allows the donor cells to be perceived by the host in a noninflammatory (nonimmunogenic) manner. Ag-SPs have been employed to prevent and treat the relapsing experimental autoimmune encephalomyelitis (EAE) model of multiple sclerosis (MS) (35) and type 1 diabetes (T1D) in the nonobese diabetic (NOD) mouse (36).

The mechanism of action of Ag-SP-induced tolerance has yet to be fully elucidated; however, the route of administration, the dosage used, the levels of costimulatory molecules expressed (the two-signal hypothesis), and the extent of T helper (Th) cell polarization and T_reg induction are all probable factors that contribute to the efficacy of the treatment. Tolerance by this strategy involves splenic marginal zone APCs, which clear the body of apoptotic cells and debris and can process the infused Ag-SPs with subsequent presentation of the coupled antigen in a tolerance-inducing manner (26, 37). The tolerogenic host APCs upregulate negative costimulatory molecules (e.g., PD-L1), with subsequent phasic tolerogenic effects on antigen-specific T cells, including anergy, deletion, and induction of T_regs (37–41). In vitro studies have suggested that engagement of the TCR in the absence of costimulation contributes to the inability to activate Th1 cell clones and leads to the induction of anergy (40). T_regs appear to have a role in antigen-coupled cell-induced tolerance, and are required for the long-term maintenance of tolerance, but are not necessary for tolerance induction (39). T cells retrieved from tolerized mice produce higher amounts of the anti-inflammatory cytokines IL-10 and transforming growth factor β (TGF-β) than do T cells from control mice, and this pattern suggests an increase in natural T_reg function. The induction of tolerance with Ag-SPs appears to work by several distinct synergistic mechanisms, increasing their therapeutic efficacy.

A recent publication presented the results of a Phase I trial in MS patients in Germany that used apoptotic ECDI-fixed peripheral blood mononuclear cells (PBMCs) pulsed with a cocktail of myelin peptides. The results illustrate the safety and efficacy of this procedure for the treatment of human autoimmune disease (42). Importantly, the mechanistic aspects of this study provided an important proof of principle that induced peripheral tolerance can be successfully employed to induce unresponsiveness in human autoreactive T cells; responses to four of the seven tolerated myelin epitopes were significantly reduced (with no effect on tetanus responses) in the four MS patients treated with >10⁹ autologous antigen-coupled PBMCs (Ag-PBMCs), and there was no effect on the nine patients receiving <5 × 10⁸ Ag-PBMCs. These studies provided the first definitive demonstration of induced tolerance to autoantigens in humans.

2.1.2. Antigen on synthetic carriers

The translational challenges associated with Ag-SP-based therapy have motivated the development of nanoparticles for antigen-specific tolerance. Nanoparticles with properties similar to apoptotic cell debris may function as an alternative antigen carrier for tolerance induction. Intravenous injection of 500-nm “nonbiodegradable” carboxylated polystyrene (PS) particles coupled with peptides prevented the onset of disease in the EAE mouse model, prevented epitope spreading, and ameliorated progression of preestablished EAE (43). Interestingly, the studies using PS nanoparticles demonstrated that particles with diameters ranging from 500 to 1,000 nm were most effective, and tolerance depended on particle uptake by the macrophage receptor with collagenous structures (MARCO) scavenger receptor (44, 45). MARCO is responsible for uptake of PS beads that have an anionically charged surface (46).

More recently, biodegradable particles made from the copolymers of lactide and glycolide (PLG) induced antigen-specific tolerance for prevention and treatment of EAE (43, 47). Administration of particles results in significantly reduced central nervous system infiltration of encephalitogenic Th1 [interferon-γ (IFN-γ)] and Th17 [IL-17a, granulocyte macrophage colony–stimulating factor (GM-CSF)] cells, as well as inflammatory monocytes/macrophages. Tolerance was most effectively induced by intravenous infusion of Ag-PLG (43, 48), although intraperitoneal and subcutaneous delivery was able to attenuate disease scores. The intravenous route likely has greater efficacy due to direct trafficking to the liver and spleen, whereas the alternative routes have distinct patterns of trafficking (39). The liver, in particular, plays a significant role in tolerance induction with nanoparticles, as splenectomized mice could be tolerized with PLG nanoparticles (D. McCarthy, W.T. Yap, C.T. Harp, W.K. Song, J. Chen, et al., manuscript in revision).

Upon intravenous delivery, particles associate with a range of APCs, including macrophages and Kupffer cells, and conventional and plasmacytoid DCs. Whereas the smallest particles reported by Getts et al. (48) were less efficient at tolerance induction, recent research with smaller particles loaded with rapamycin were employed to promote tolerance via intravenous delivery (49). Similarly, tolerance induction was demonstrated using gold nanoparticles loaded with 2-(1′H-indole-3′-carbonyl)-thiazole-4-carboxylic acid methyl ester (ITE, a ligand for the aryl hydrocarbon receptor) and antigen (50). Differences in cellular association between small and large particles may underlie the need for a pharmaceutical agent along with the smaller particles (49, 50). More recently, nanoparticles were employed to target liver sinusoidal endothelial cells (51), which can induce CD4⁺Foxp3⁺ T_regs (52). This nanoparticle-based antigen delivery prevented the onset of clinical EAE and, when delivered therapeutically, improved clinical scores following a single dose. The T_reg frequencies in the spleens of mice treated with autoantigen peptide–loaded nanoparticles were significantly higher than those in vehicle-treated mice. Moreover, nanoparticle-mediated disease control was abrogated after T_reg depletion by repeated administration of T_reg-depleting anti-CD25 antibody. Other groups have also investigated particles for the induction of T_regs in vitro and in vivo by delivering combinations of immune-modifying agents such as IL-2, TGF-β, or rapamycin (53, 54). Understanding the cell types associated with particles in vivo will enable the development of more effective tolerizing therapies that focus on optimizing cell-specific tolerogenic responses.

3.1.1. Breadth of antigens

The primary antigens that trigger the host rejection immune response are the MHCs; in humans, these are called human leukocyte antigens (HLAs). The HLA genes exhibit extreme polymorphism, and thousands of new alleles have been and are continuing to be identified. However, the immunogenicity of HLA mismatches has recently been suggested to stem from individual alloreactive “determinants” or “epitopes” within each HLA antigen (99). Every HLA antigen has a unique set of such epitopes, although many are shared between different HLA antigens. Consequently, each HLA mismatch, in essence, could be viewed as a set of multiple epitope mismatches. In any given donor–recipient pair, the number of HLA mismatches multiplied by the number of different epitopes in these HLA antigens results in a large number of potentially immunogenic epitope mismatches. To further complicate the situation, as evidenced in rejection in HLA–identically matched transplants, non-HLA or minor histocompatibility antigens (mHAs) have also been implicated in eliciting strong cellular immune responses. Although the Y chromosome–encoded male-specific antigens were the first identified mHAs, based on the known abundance of functional variants in the human genome and recent rapid genomic advances, the number of mHA mismatches between any given donor–recipient pair is expected to be large (100). Two important aspects of the potentially large numbers of HLA and mHA mismatches should be considered when assessing their importance in transplant rejection and tolerance. First, it is likely that different mismatches elicit immunogenicity of a wide range of strength, and the same mismatch may elicit different immunogenicity depending on recipient antigen processing and presenting HLAs. Second, when considering antigen-specific tolerance strategies (as detailed in Section 3.2, below), engineered tolerance to one epitope may result in cotolerance (bystander regulation) to other epitopes that are expressed by the same cells, a situation that has previously been described as linked suppression (101). The latter possibility may be exploited to reduce the complexity of the target transplant antigens.

3.1.2. Redundant effector pathways

Transplant immunity is uniquely robust because it can be triggered by several parallel antigen presentation pathways (97): direct antigen presentation by donor-derived APCs presenting donor HLAs, indirect antigen presentation by recipient-derived APCs presenting processed donor HLA peptides, and semidirect antigen presentation by recipient-derived APCs that have acquired and now present intact donor HLAs. The subsequent effector mechanisms triggered by these antigen presentation pathways are also varied. Whereas classical Th1 CD4⁺ T cells and cytotoxic CD8 T cells are thought to be mainly responsible for rejection, recent studies have implicated a whole spectrum of other effector cells in this process, including Th2 cells, Th17 cells, memory CD8 T cells, and cells of the innate immune system such as monocytes and natural killer cells. Which effector pathway(s) dominates in any given rejection process varies depending on the specific tissue/organ transplanted and the host immune composition (e.g., microbiota, presence or absence of other inflammatory signals). In addition, suppression of one effector pathway may lead to the induction of an alternative effector pathway to promote rejection (102). The challenge resulting from this redundancy is that a robust tolerance strategy will likely need to effectively control multiple pathways. At the same time, effective tolerance approaches will likely need to be personalized on the basis of best-predicted effector pathways involved in a given patient and for the transplant of a specific tissue.

3.1.3. Prior sensitization

Transplant recipients are frequently sensitized to alloantigens because of prior blood transfusions, pregnancies, and/or transplantation. Sensitized recipients may manifest preexisting anti-HLA antibodies, which may fix complement and mediate cytotoxicity upon binding to the recognized HLA antigens on the transplanted organ, leading to hyperacute rejection of the transplanted organ. This situation can now be effectively avoided by ensuring pretransplant elimination of such antibodies by desensitization (103), a process that usually involves plasmapheresis. However, in addition to such humoral sensitization, cellular sensitization is also a significant barrier. Allospecific memory T cells can mount robust antidonor responses even with minimal costimulation signals, and memory B cells may be capable of rapidly developing into antibody-secreting plasma cells even in the absence of T cell help (104, 105). These “shortcuts” frequently evade and nullify conventional tolerance mechanisms, and may additionally turn a donor cell–based tolerance therapy into an exacerbating event. Consequently, the design of tolerance therapy in presensitized recipients will need to (a) eliminate the possibility of the tolerogen turning into an immunogen and (b) effectively tolerize memory T and B cells, the latter of which requires an improved understanding of the precise cellular pathways involved in memory T and B cell activation and their effector functions.

3.2.1. Mixed chimerism

Mixed chimerism is a state in which donor-derived hematopoietic cells cocirculate with recipient cells without being rejected. When mixed chimerism is achieved, recipients become tolerant to the donor MHC antigens and consequently readily accept a transplanted allograft from the same donor. Induction of mixed chimerism with donor hematopoietic stem cell infusion has recently been achieved in clinical kidney transplantation in humans (106–110). The number of donor-derived hematopoietic cells could be high (>1% of total leukocytes, known as macrochimerism) and readily detectable by conventional methods such as fluorescence-activated cell sorting (FACS), or it could be low (<1% of total leukocytes, known as microchimerism) and detectable only by highly sensitive methods such as polymerase chain reaction (PCR). The induced chimerism could be durable or transient. The predominant mechanism thought to underlie tolerance by macrochimerism is deletional, in which high numbers of donor APCs continuously delete newly developing alloreactive T and B cells in primary lymphoid organs such as the thymus and the bone marrow. Thus, the stability of such tolerance relies on the persistence of donor chimerism. Conversely, the predominant mechanism thought to underlie tolerance by microchimerism is regulatory, in which T_regs are present in peripheral tissues. Clinically, mixed chimerism has been achieved in humans by several existing protocols involving total lymphocyte irradiation (TLI) or localized thymic irradiation, as well as various cytoreductive therapies (e.g., rabbit antithymocyte globulin, cyclophosphamide, anti-CD2 monoclonal antibody, fludarabine), in combination with donor hematopoietic stem cell infusion and, in one case, novel tolerogenic “facilitating cells.” Such mixed chimerism has been achieved in both HLA-matched and HLA-mismatched donor–recipient pairs. Although these groundbreaking successes have led to immunosuppression-free graft survival in a substantial percentage of recipients, their immune reconstitution and long-term immune competence, the stability of tolerance in macro- versus microchimerism, and the long-term risk of graft-versus-host disease (GVHD) remain to be defined.

3.2.2. Donor negative vaccines

The concept of donor negative vaccines stems from the observation that T cell activation requires two signals (Figure 1). Experimental protocols blocking the CD28/CD80/CD86 axis or the CD40/CD40L axis (signal 2), in combination with providing signal 1 (e.g., donor-specific transfusion), can successfully induce long-term tolerance. However, this approach has yet to be translated to clinical practice. Treatment with the fusion protein CTLA-4–Ig (belatacept) in clinical kidney transplantation to block the CD28/CD80/CD86 pathway is associated with a higher frequency of acute rejection, although long-term graft function is not impaired (111). This adverse effect may be due to the escape of memory T cells from the need for positive costimulation, a reduction in the number of the suppressive T_regs due to their dependence on positive costimulation, and/or blockade of negative costimulation by this agent. Treatment with anti-CD154 to block the CD40/CD40L pathway in humans resulted in the development of unexpected thrombotic events due to the expression of CD154 on human platelets (112). With the development of newer costimulation blockade agents that have improved safety profiles in humans, there is a renewed interest in further exploration of the two-signal hypothesis for transplant tolerance induction in the clinic.

3.2.3. Suppressor cells

Several suppressor cell populations have been demonstrated to play important roles in mitigating transplant rejection. The best-studied suppressor cell population is the CD4⁺CD25⁺Foxp3⁺ T_regs. These cells are critical for the induction and maintenance of peripheral tolerance. In transplantation, they migrate to the graft and the graft-draining lymph nodes, where they suppress donor-stimulated effector T cell function (113–115). Since the first report of human T_reg expansion, many investigators have experimented with protocols for non-specific or donor-specific expansion of natural T_regs, as well as with protocols for conversion and expansion of induced T_regs. The multicenter international ONE study, involving eight clinical centers in Europe and the United States, is currently under way to test the safety and feasibility of seven different T_reg populations in living donor kidney transplantation (116). With such concerted efforts, defining the optimal source, type, number, in vivo stability, and efficacy of T_reg therapy will hopefully occur soon.

The second suppressor cell population is mesenchymal stem cells (MSCs). MSCs are multipotent progenitor cells that inhibit innate immunity as well as adaptive immune responses associated with transplant rejection. They suppress the activation and proliferation of both naïve and effector T cells and promote the expansion of T_regs. MSCs were recently demonstrated to have the unique ability to control the expansion of memory T cells, a major barrier in clinical transplantation, as described above. Both autologous and allogeneic MSCs have been studied in animal models in organ or cell transplantation and have shown potential tolerogenic characteristics. The first inhuman study of MSCs in kidney transplantation was published by Tan et al. (117) in 2012; in this experiment, infusion of autologous MSCs at the time of surgery and 10 days later demonstrated superior early graft function, reduced the frequency of acute rejection, and allowed reduction of maintenance calcineurin inhibitor in comparison to conventional therapy. However, clinical application of allogeneic and/or third-party MSCs awaits further studies.

The third suppressor cell population is myeloid-derived suppressor cells (MDSCs). These cells are typically CD11b⁺Gr1⁺ in mouse models of transplantation and are considerably less well defined in human recipients of transplantation; they were only recently identified in renal transplant patients as CD11b⁺CD33⁺HLA-DR⁻ cells with variable levels of expression of CD14 and CD16, underscoring their phenotypic heterogeneity (118–121). They mediate suppression by several mechanisms, including inhibition of cytotoxic T cell responses by generation of reactive oxygen species and IL-10, depletion of L-arginine via arginase 1 and inducible nitric oxide synthase, depletion of tryptophan via indoleamine 2,3-dioxygenase (IDO), and promotion of T_regs via IFN-γ-dependent pathways. The use of MDSC adoptive transfer cell therapy in modulating allograft rejection has, however, been demonstrated only in animal models of transplantation. Clinical applications that harness the potential of this suppressor cell population will depend on the availability of in vitro protocols that can reproducibly generate well-defined MDSC populations that suppress human alloimmune responses.

3.4.1. ECDI-treated cells

Within the context of transplant tolerance, intravenous delivery of donor SPs chemically treated with ECDI (ECDI-SPs) have shown robust efficacy in allogeneic islet, heart, and xenogeneic islet transplantation in mouse models by taking advantage of mechanisms similar to those of Ag-SP treatment of autoimmunity (37, 137, 138). However, no additional coupling of antigens is necessary, because donor cells contain the full spectrum of allogeneic or xenogeneic antigens. This tolerance strategy is donor specific (139). This approach is also flexible because B cells of donor origin can be expanded in vitro and subsequently treated with ECDI to induce tolerance. Additionally, the lysate from allogeneic donor cells can be conjugated to syngeneic SPs with ECDI to induce tolerance with no loss of efficacy (140). These findings support the study of the clinical applicability of this approach.

Tolerance mediated by donor ECDI-SPs depends on low APC expression of positive costimulatory molecules such as CD80 and CD86 and enhanced expression of negative costimulatory molecules such as PD-L1 and PD-L2 (141, 142). The immunoregulatory cytokine milieu induced by recognition of apoptotic cells is likely the reason underlying the establishment of an environment wherein negative costimulation is the favored outcome. Supporting this hypothesis are the findings that splenic macrophages express IL-10 following infusion of ECDI-treated antigen coupled leukocytes and that inhibition of IL-10 following infusion of ECDI-SPs prevents tolerance in the context of both autoimmunity and allogeneic islet transplantation (39, 137, 143). A second cytokine implicated in donor ECDI-SP-induced tolerance is IFN-γ. This cytokine is initially produced by CD4 T cells that indirectly recognize donor antigens presented by self-APCs, and it later mediates depletion of such donor-specific T cells and establishment of transplant tolerance (144). Recent experiments indicate that another soluble factor, IDO, can be induced by infusion of apoptotic donor ECDI-SPs (143, 145). The critical role of this factor in transplanting tolerance induced by ECDI-SPs is demonstrated by the abolishment of tolerance when an inhibitor of IDO, 1-methyl-D-tryptophan, is administered at the time of ECDI-SP infusion (145).

Two regulatory cell populations are induced by infusion of donor ECDI-SPs and play a critical role in mediating its tolerogenic effects. In vivo, T_regs are preferentially expanded in frequency in the secondary lymphoid organs and grafts of ECDI-SP-tolerized transplant recipients, and depletion of CD25⁺ T_regs at the time of donor ECDI-SP treatment prevents the establishment of transplant tolerance. Furthermore, the requirement for TGF-β at the time of tolerance induction by donor ECDI-SPs supports the idea that these T_regs are induced from naïve T cells rather than expanded from existing T_regs (143). Another suppressor cell population induced by donor ECDI-SPs is the MDSCs. Treatment with allo-ECDI-SPs induced a splenic population of MDSCs that produced significant levels of IFN-γ-dependent IDO and suppressed CD8⁺ T cell proliferation when compared with control mice. Moreover, MDSCs were present in the cardiac allograft and suppressed the infiltration of CD8⁺ T cells and other effectors. MDSCs present in the graft also produced IL-10 and recruited T_regs in a CCL4-dependent manner, and depletion of MDSCs restored CD8⁺ T cell infiltration and graft rejection (121).

3.4.2. Nanoparticles

The need for many fresh donor cells in the manufacturing of tolerance cell products (i.e., ECDI-SPs) may be logistically challenging and may also introduce variability and opportunities for pathogen transmission. These issues could be overcome through the use of PLG particles as carriers for delivering donor antigens. PLG particles used for autoimmune tolerance have been adapted to the delivery of alloantigens (146). Given the antigenic diversity, allogeneic tolerance was expected to be inherently more challenging than tolerance to autoimmunity (43, 47, 48). However, our data show that PLG particles linked to solubilized donor antigens (PLG-dAg) can promote long-term islet allograft function (146). Interestingly, we now know that, in addition to modifying adaptive immunity, PLG particles can also modify innate immunity by targeting inflammatory monocytes via MARCO, leading to sequestration in the spleen of monocytes that are incapable of migrating to injured tissues to induce inflammation (48).

We have developed nanoparticles loaded with donor cell lysates and tested their ability to provide long-term protection to islet allografts in a BALB/c → B6 transplant model (146). Donor cell lysate containing donor antigens was coupled to PLG particles through ECDI chemistry to generate PLG-dAg. Delivery of PLG-dAg alone provided a slight advantage to graft survival compared with treatment with empty PLG particles. This is in sharp contrast to the high efficiency of tolerance induction to autoimmunity by peptide-coupled PLG particles (43, 47), highlighting the differences in alloimmune versus autoimmune tolerance by antigen-delivering PLG particles. However, when PLG-dAg particles were combined with a low-dose (0.1 mg/kg) short course (4 days total from day −2 to day +1) of rapamycin, significant long-term graft protection was observed in ~60% of recipients (146). These studies demonstrate the potential of PLG particles for donor antigen–negative vaccination for allogeneic tolerance induction.

Antigen-coupled nanoparticles are likely to be more versatile than ECDI-SPs because they can be engineered to potentially target different host cell populations and consequently elicit a wide spectrum of desired in vivo effects. For the induction of transplantation tolerance using PLG-dAg, our early studies demonstrated that PLG-dAg displayed a similar biodistribution profile to ECDI-SP and accumulated primarily in the liver, spleen, and lungs following intravenous injection (146). Within the spleen, PLG-dAg appears to be internalized by F4/80⁺ macrophages, CD11c⁺CD11b⁺ cells, CD11c⁺B220⁺ cells, CD11c⁺CD8α⁺ DC subsets, and B220⁺ B cells, again similar to internalization of ECDI-SPs (146). Importantly, ECDI-PLG-dAg modulates T cells with indirect donor specificity through initial clonal expansion followed by clonal contraction. However, in contrast to ECDI-SPs, PLG-dAg does not seem to anergize T cells with direct donor specificity (146). These findings provide us with initial ideas for improving PLG designs in order to maximize tolerance efficacy, including the route of administration, properties of the nanoparticle platform (size and surface charge), and codelivery immunotherapeutics.

3.4.3. Codelivery of immunotherapeutics

Codelivery of immunotherapeutics along with donor antigens using nanoparticles is an attractive engineering option that may result in significant enhancement of tolerance efficacy. One possibility is codelivery of immunosuppressive/immunomodulatory agents by nanoparticles that may further attenuate the response of T cells interacting with the targeted APCs internalizing the nanoparticles via donor-specific TCRs. Choices include encapsulation of pharmacological agents such as rapamycin and suppressive cytokines such as IL-10, TGF-β, and IDO. Another option is the engineering of tolerogenic ligands into the nanoparticles. For example, incorporating surface phosphatidylserine may promote recognition of the nanoparticles as “apoptotic,” thereby facilitating uptake of the nanoparticles by host phagocytes via specific pathways that lead to “tolerogenic” antigen presentation (147). Other tolerogenic ligands include the cationic polymer polyamine polyethylenimine and ITE, which exploit the immunomodulatory pathways of IDO and aryl hydrocarbon receptor in interacting DCs (148, 149). Lastly, targeted delivery of the donor antigen cargo only to specialized DC populations may preferentially amplify the desired tolerant response. One such DC target is DEC205⁺ DCs. PLG particles functionalized with anti-DEC205 antibodies have been reported to cross-link the DEC205 receptor, resulting in enhanced production of IL-10 (150).