Mixed chimerism and split tolerance

Abstract

Establishing hematopoietic mixed chimerism can lead to donor-specific tolerance to transplanted organs and may eliminate the need for long-term immunosuppressive therapy, while also preventing chronic rejection. In this review, we discuss central and peripheral mechanisms of chimerism induced tolerance. However, even in the long-lasting presence of a donor organ or donor hematopoietic cells, some allogeneic tissues from the same donor can be rejected; a phenomenon known as split tolerance. With the current goal of creating mixed chimeras using clinically feasible amounts of donor bone marrow and with minimal conditioning, split tolerance may become more prevalent and its mechanisms need to be explored. Some predisposing factors that may increase the likelihood of split tolerance are immunogenicity of the graft, certain donor-recipient combinations, prior sensitization, location and type of graft and minimal conditioning chimerism induction protocols. Additionally, split tolerance may occur due to a differential susceptibility of various types of tissues to rejection. The mechanisms involved in a tissue’s differential susceptibility to rejection include the presence of polymorphic tissue-specific antigens and variable sensitivity to indirect pathway effector mechanisms. Finally, we review the clinical attempts at allograft tolerance through the induction of chimerism; studies that are revealing the complex relationship between chimerism and tolerance. This relationship often displays split tolerance, and further research into its mechanisms is warranted.

Chimerism and Tolerance

Induction of donor-specific tolerance to transplanted organs or tissues is one of only a few approaches with the potential to eliminate the need for long-term immunosuppressive therapy, while also preventing chronic rejection. Establishing hematopoietic chimerism is one such method of inducing donor-specific tolerance. Chimerism was first associated with tolerance in the observations of Owen in which fraternal cattle twins were shown to be natural chimeras and therefore operationally tolerant of one another.¹ It is also likely that the demonstration of acquired tolerance induced by Billingham et al. through the injection of “testis, kidney and splenic tissue” into fetal mice involved the creation of hematopoietic chimerism.²

Tolerance mechanisms in mixed chimeras

Tolerance in mixed chimerism involves both central and peripheral mechanisms. After bone marrow transplantation, donor stem cells migrate to and proliferate in the host bone marrow compartment.³ Donor stem cell hematopoiesis leads to mixed chimerism and populates the thymus with the hematopoietic cells involved in negative selection. In the thymus, donor and recipient antigen presenting cells will then eliminate both donor-reactive and host-reactive T cells.^4-6 After transplant, donor antigens can be presented to anti-donor T cells ‘directly’ on the major histocompatibility complex (MHC) of donor cells, or ‘indirectly’ when processed and presented on the MHC of recipient cells. Importantly, both sets of anti-donor T cells, i.e., those with direct anti-donor specificity and those with indirect anti-donor specificity, can be made tolerant in the thymus.^7,8 Thus, chimerism takes advantage of central tolerance, a robust form of tolerance in a manner similar to how the immune system evolved to eliminate most self-reactive responses.⁹ However, as discussed in more detail further on, it is unlikely all donor antigens reach the thymus to induce central tolerance.

Central tolerance

Central tolerance is believed to be the dominant mechanism of tolerance in mixed chimerism. However, although many experiments demonstrate central (deletional) tolerance is occurring in chimeras, few experiments have actually tested whether it is actually required for chimerism-induced tolerance to alloantigens. Evidence for the occurrence of central tolerance in non-myeloablative mixed chimerism protocols comes from multiple experimental models. Early experiments demonstrating central deletion in chimerism used donor-recipient mouse combinations that differ in MHC class II I-E expression, thereby allowing the tracking of superantigen reactive T cells that express certain Vβ families. In this manner, it has been shown that donor reactive mature T cells are centrally deleted soon after the induction of chimerism.⁶ These results were confirmed in a CD8 T‑cell receptor transgenic mouse model, made chimeric with MHC mismatched bone marrow; the transgenic CD8 T cells were deleted in the thymus.⁵ Since antigen-presenting cells are potent mediators of negative selection in the thymus,^10,11 indirect evidence for negative selection in mixed chimeras comes from the association between donor MHC class II positive cells in the thymus and tolerance to skin^5,12 or kidney grafts.¹³

Although the above-mentioned studies demonstrate central deletion can indeed occur in the setting of chimerism, none provide evidence that central tolerance is actually needed for mixed chimerism to induce tolerance. Similarly, the fact that thymectomy of chimeras prevents a loss of tolerance upon depletion of the chimeric donor cells¹⁴ is not evidence that central tolerance was required. There is a fundamental difference between demonstrating that central tolerance occurs in the setting of chimerism and demonstrating it is required for chimerism to induce tolerance. Whether central tolerance is required may be at the heart of understanding why one cellular transplant (hematopoietic cells) generates tolerance but another cellular transplant (e.g., pancreatic islets) does not. Or, in other words, is there something unique about hematopoietic cells that allow them to generate tolerance, or is it simply the fact that they are able to migrate to the thymus and induce central tolerance? Our group has investigated the ability of donor tissue transplants vs. donor hematopoietic cells to naturally induce tolerance, i.e., without drug treatment. These studies were aimed at defining the ‘rules’ of self-tolerance generation, using graft alloantigens as a model. Using this model we could ask, to be treated as self (i.e., induce natural tolerance), does an (allo)antigen present in a recipient need to be present systemically or can it be localized in the periphery? To provide a setting where such natural tolerance has optimal conditions to take hold, we transplanted donor tissue into recipients before the recipient’s immune system had begun to develop. In one example, we grafted male skin onto female immunodeficient (athymic) mice followed several months later by a female thymus graft to allow T‑cell generation to begin in the recipients.¹⁵ These experiments showed passenger lymphocyte chimerism was required for natural tolerance and that it can induce central tolerance. In addition, these experiments also demonstrated chimerism was unable to generate natural peripheral tolerance in adoptively transferred mature splenic T cells (mature T cells were instead immunized), suggesting central deletion was the dominant mechanism. However, although supportive, this study also could not prove that the central tolerance was necessary for the chimerism to induce tolerance. For example, these studies were done in highly lymphopenic immunodeficient mice, and the homeostatic activation that occurs in this setting may have blocked peripheral tolerance induction.¹⁶ However, more recently we have shown that chimerism-generating passenger lymphocytes from an islet transplant increase the rate of rejection in wild-type immunocompetent recipients, supporting the contention that chimerism is immunogenic in the periphery;¹⁷ these studies also showed that donor dendritic cells are not the only immunogenic passenger cells in a transplant.

Despite the above studies, there remains an additional issue that requires resolution before making the conclusion that central tolerance is necessary for chimerism to induce natural tolerance. It might be that only newly generated T cells have the appropriate programming to become naturally tolerant upon encountering donor chimerism in the periphery. In the above studies, the immunization of mature T cells may have been due to their functionally mature programming as opposed to more immature recent thymic emigrants,^18,19 which might be more susceptible to tolerance. Perhaps newly generated T cells may be tolerized by encounter with donor chimerism in the periphery, with no requirement for central tolerance. Moreover, the passenger lymphocyte chimerism in the above studies was mainly T cells, which lack MHC class II expression; expression of MHC class II might be a prerequisite for donor cells to induce peripheral tolerance. Only recently was the requirement for central tolerance in-chimerism specifically examined.²⁰ In this study, we designed experiments to directly test whether endogenous, newly produced T cells exiting the thymus and encountering the chimerism only in the periphery become tolerized or immunized by the donor chimerism. Again, we placed donor cells into immunodeficient recipients prior to immune system development of the recipients. We found that chimerism (donor-derived peritoneal leukocytes) would be accepted only if the donor leukocytes were able to establish systemic chimerism, but not when present, in a localized fashion, solely in the periphery. This experiment confirms that newly generated T cells do not become peripherally tolerant of donor leukocytes (for an exception to this rule see ref. 16), and argues that any potential increased susceptibility of newly generated T cells to tolerance induction²¹ is insufficient to allow peripheral tolerance to be established against the highly mismatched chimeric donor cells. Additionally, the chimerism in these experiments was composed mostly of B cells that express both MHC I and II. Therefore, peripheral mechanisms alone appear insufficient to generate tolerance to MHC mismatched chimeric cells, as tolerance did not develop, even in newly generated T cells, unless the chimerism was present systemically. However, these conclusions are also based on studies in a lymphopenic model, and it remains possible that in a non-lymphopenic setting, newly generated T cells may demonstrate an ability to be tolerized in the periphery by chimerism without any need for central tolerance to the donor.

Taken together, our studies indicated that, with the exception of very weakly mismatched transplants,²⁰ natural tolerance to donor tissues transplanted before recipient immune system development cannot be established without donor cells reaching the thymus. This finding suggests a parsimonious solution to the long-standing paradox that transplants given to recipients before immune system development (e.g., grafts given to a fetus) trigger rejection in some cases and tolerance in others. Of historical interest, the key requirement for central mechanisms in tolerance of such ‘pre-immunocompetence transplants,’ and tolerance in general, is predicted by Lederberg’s theory.²² This 1959 theory, often misattributed to Burnet, who even years later had not appreciated the advance made by Lederberg’s postulate that it is the maturity of the lymphocyte and not the maturity of the animal that is the key solution to tolerance.

Another important example of the critical role of chimerism in tolerance of foreign cells that appear early in life has been elucidated in studies of tolerance to non-inherited maternal antigens (NIMA). This tolerance, caused by maternal cell microchimerism in the fetus, involves a fetal regulatory T‑cell (Treg) cell response.^23-25 The studies discussed above suggest that tolerance to NIMA may require central tolerance of the antigens of the cells that set up the microchimerism (stem cells). Furthermore, the regulatory mechanisms observed may be in response to the progeny of the maternal microchimerism that differentiates to express a distinct set of antigens.²⁶ Such a scenario predicts that the depletion of Tregs would reveal immune reactivity toward the differentiated progeny of maternal microchimerism (e.g., maternal DCs) without a response to the maternal stem cells that initially established the microchimerism. Whether peripheral tolerance alone is sufficient for this state of natural tolerance to allogeneic cells has not yet been determined. Future investigations will be important to determine if tolerance to NIMA requires central tolerance or is instead due solely to peripheral mechanisms. These peripheral mechanisms may include immune escape as a result of the relatively tiny number of cells introduced, low MHC expression, and/or residence in an ‘immune privileged’ niche, acting together with the aforementioned Treg responses.

Peripheral tolerance

Many of the experiments suggesting the key role of central mechanisms in chimerism-induced tolerance have used T‑cell depleting induction protocols;^5,6 therefore, peripheral mechanisms of tolerance were thought to play only a minor role. Although T‑cell depleting protocols are non-myeloablative, the drive toward developing clinically relevant mixed chimerism induction protocols began shifting away from T‑cell depletion and toward co-stimulatory blockade. This shift brought with it the notion that central deletion cannot account for the tolerization of pre-existing donor-reactive T cells, and peripheral mechanisms must also be present. Subsequently, both therapy-induced deletional and non-deletional peripheral tolerance mechanisms involved in mixed chimerism were identified. Tracking T cells with Vβ reactive to donor superantigen, in a co-stimulatory based chimerism induction protocol, it was found that directly donor reactive T cells were rapidly deleted in the periphery.²⁷ To ensure total donor reactive T‑cell numbers were not diluted due to contributions from the recently centrally tolerized T‑cell repertoire, thymectomized chimeras were shown also to have deleted a subset of donor reactive peripheral T cells. These results were supported by another study that showed peripheral deletion is due to the co-stimulatory blockade and not signaling through the co-stimulatory blocking antibody.²⁸ This study also demonstrated the presence of therapy-induced non-deletional peripheral tolerance mechanisms because tolerance develops in this mixed chimerism model before the elimination of donor-reactive cells. T‑cell anergy is rapidly seen upon induction of chimerism with co-stimulatory blockade.^29,30

In contrast to anergy, the identification of regulatory mechanisms of tolerance involved in co-stimulatory blockade induced chimerism has been challenging and has sometimes provided conflicting conclusions. For example, some studies have found recipient CD4 cells are required for the induction of chimerism.^12,30,31 These results, combined with in vitro³¹ and in vivo^32-34 suppression experiments may point toward a role for regulatory T cells in establishing the chimerism-induced tolerance. Additionally, in canine chimerism models of lung or vascularized composite tissue allografts, the frequency of T regs is increased in blood and draining lymph nodes, respectively.^35,36 However, in our own and others’ experiments,^34,37,38 depletion of CD4 T cells at the time of bone marrow transplantation has enhanced sustainable chimerism. Further arguments against regulatory T‑cell involvement, at least in maintenance of tolerance, comes in part from experiments where the depletion of CD4 cells months after bone marrow transplant does not affect tolerance.¹² Additionally, adoptively transferring splenocytes from chimeric mice into immunodeficient hosts could not induce tolerance to donor skin grafts and injection of naive recipient splenocytes into stable chimeric mice brakes tolerance.²⁹ In another experiment, after BMT some mixed chimeras did not develop T cells; they only developed peripheral granulocytes, B and NK cells. In contrast to chimeric mice that did generate T cells, the chimeras without T cells were not tolerant of donor skin grafts. Even when sorted CD4 CD25 T cells from tolerant animals were transferred into the chimeras that did not generate their own T cells, tolerance toward skin grafts could not be induced.³⁸ However, in some patients who received non-myeloablative conditioning and a renal allograft,³⁹ the recovery of CD4 T‑cell populations was associated with higher frequencies of CD25^high CD127^- FoxP3⁺.⁴⁰ This high frequency was associated with suppression of anti-donor reactivity in 2/4 patients at 6–12 mo post-transplant, but suppressive function was lost after one year.

Although deletion, anergy and regulation may play roles in peripheral tolerance in chimeras, the exact contributions of each may differ between model systems.⁴¹ Alternatively, peripheral tolerance may have an important function in the immediate post-bone marrow transplant period, but play a progressively smaller role as central tolerance becomes established.

Split Tolerance

Even in the long-lasting presence of mixed chimerism, some allogeneic tissues from the same donor can be rejected; a phenomenon known as split tolerance.^37,42-45 Experimentally, split tolerance occurred when the first mouse mixed chimeras were created using lethal irradiation and bone marrow reconstitution with mixed syngeneic and allogeneic cells.⁴² After being identified in lethally irradiated mixed chimeric mice, split tolerance was subsequently seen in non-myeloablative mixed chimeras.^37,44,45 Beyond mice, split tolerance has also been seen in large animal models^46-48 and human clinical trials^39,49,50 of mixed chimerism. By organ, split tolerance has been seen in large animals toward skin,^47,48 kidney,⁴⁶ heart⁵¹ and hematopoietic cells.^39,52,53 With the current goal of creating mixed chimeras using clinically feasible amounts of donor bone marrow and with minimal conditioning, split tolerance may become more prevalent and its mechanisms need to be defined.

A number of factors may be associated with an increased likelihood of split tolerance developing in mixed chimeras (Fig. 1). Predisposing factors may include immunogenicity of the graft, certain donor-recipient combinations, prior sensitization, location and type of graft (cellular or solid) and minimal conditioning chimerism induction protocols. Additionally, through chimerism and non-chimerism experiments, a number of explanations have been proposed to explain the differential susceptibility to rejection of various types of tissues. These explanations include the presence of polymorphic tissue specific antigens and variable effectiveness of indirect effector mechanisms to reject certain allografts.

Figure 1.

Predisposing factors and potential mechanisms for the generation of split tolerance

Increasing the probability of split tolerance

Immunogenicity of graft

The inherent immunogenicity of the tissue transplanted to mixed chimeras may predispose the graft split tolerance. Once chimerism is established, it is possible that a subsequent, highly immunogenic, donor tissue graft may be rejected (especially if allelic tissue-specific antigens are present). Peripheral tolerance mechanisms operating in mixed chimeras may become overwhelmed in the presence of considerable activating signals and antigen presentation, leading to split tolerance. For example, skin possesses properties of potent antigen presentation by Langerhans cells or dermal DCs and high concentrations of extracellular glycoproteins that may facilitate T‑cell activation. However, this latter hypothesis does not explain why minor antigen matching prevents rejection of skin grafts in chimeras.^37,41,42 Indeed, the difficulty in skin graft acceptance in mixed chimerism is similar to that seen in non-chimerism experiments, where a mismatch for a single antigen is sufficient to cause skin rejection.⁵⁴ However, being mismatched for a single antigen is not sufficient for heart rejection, and yet there can be split tolerance to heart.⁵⁵ Perhaps, non-vascularized allografts (e.g., skin) may be more susceptible to immune destruction than vascularized grafts (e.g., heart). Skin grafts are subject to ischemic insults, and continuous exposure to microbial flora, that may lead to inflammation and necrosis. This non-specific inflammation may make such non-vascularized grafts more susceptible to subsequent immunological destruction.⁵⁶ Again, peripheral tolerance mechanisms may become overwhelmed in the presence of significant inflammatory signals leading to split tolerance to the non-vascularized allograft. However, despite the theoretical difference in immunity based on vascularity, skin grafts were rejected much more rapidly than non-vascularized heart grafts in a MHC-mismatched donor-recipient combination.⁵⁷ Additionally, vascularized skin grafts are rejected in an identical fashion to full-thickness skin grafts.⁵⁸ Although it is debatable as to the exact property of skin that increases its immunogenicity, conceivably, non-specific inflammation and potent antigen presenting properties may predispose it to rejection. However, since split tolerance can occur to other tissues that, unlike skin, do not possess high immunogenicity, the aforementioned factors are not universal and cannot account for split tolerance toward all allografts.

Minimal conditioning

In 1955, the first success at experimentally creating chimerism was done on mice with lethal dose irradiation in order to create an immunological ‘clean slate.’⁵⁹ The total destruction of the host hematological system allows the complete reconstitution of this system by donor bone marrow cells (full chimerism), and the subsequent acceptance of donor skin transplants. However, there are some drawbacks with inducing full chimerism. First, although often successful at inducing chimerism and tolerance to donor tissues, it had already been established that transplantation of tissues that contain significant numbers of immunologically active cells carries the risk of graft-vs.-host disease (GVHD).^60,61 GVHD occurs when donor immunocompetent cells react against host antigens and is a major concern in transplant protocols that create full chimerism, especially when involving MHC mismatches. Second, full chimeras may be relatively immunodeficient, potentially as a consequence of donor T cells maturing in a MHC mismatched host thymus.^62,63 Although this immunodeficiency does not occur in humans, where at least partial MHC matching is performed. Last, the myeloablative conditioning involved in the creation of full chimeras is toxic and is associated with high morbidity, prohibiting its use outside of treatments for malignancies. For chimerism to be widely applicable in transplantation protocols, the risks of GVHD must be minimized and milder conditioning regimens implemented.

Mixed chimeras, organisms with a variable balance of donor and recipient hematopoietic cells, are more clinically favorable than full chimeras. Similar to full chimerism induction protocols, mixed chimeras were created using lethal irradiation; the difference being that the immune system was reconstituted with a mixture of syngeneic and allogeneic bone marrow.⁶⁴ These mixed chimeras were tolerant to donor skin grafts and rejected third party skin grafts. However, the donor recipient combination employed in this study was matched for all minor antigens, a situation that artificially avoids split tolerance and yet is not a realistic approach for the clinic. Importantly, the mice were immunocompetent as suggested by their longer survival when compared with full chimeras.⁴² Since the host hematopoietic system remains intact in mixed chimeras, antigen-presenting cells with host MHC are able to interact with positively select T cells, potentially contributing to a more competent immune system than full chimeras.⁶⁵

Although efficient at inducing mixed chimerism, the toxicities and undesirable side effects of myeloablative protocols led to the development of non-myeloablative chimerism induction protocols. For example, one successful non-myeloablative protocol to establish mixed chimerism used T‑cell depleting monoclonal antibodies, low dose total body irradiation (TBI) and additional radiation to the thymus.⁶⁶ In addition to low dose TBI and T‑cell depletion, further efforts to minimize host conditioning in order to translate mixed chimerism from experimental to clinical settings have involved the implementation of donor-specific transfusion, co-stimulatory blockade, and short-term use of immunosuppressive medications.

Split tolerance may be more likely to occur in minimal conditioning (non-myeloablative) mixed chimerism induction protocols that leave more of the recipient T‑cell compartment intact. In such circumstances, successful induction of tolerance becomes more dependent on peripheral tolerance, for both direct and indirect donor-reactive T cells. However, with many different chimerism induction protocols, and the relative rarity of experimental split tolerance, it is difficult to isolate the effects each conditioning agent-has on the development of split tolerance.

Mechanisms of split tolerance

Tissue-specific alloantigens

In many studies, mixed chimeras maintained donor hematopoietic cells but rejected skin transplants,^37,42-45 the cause of which was likely immunity toward polymorphic tissue-specific alloantigens expressed by donor skin but absent from donor bone marrow cells.^37,42,67-70 Therefore, although chimerism can induce tolerance toward hematopoietic cells, that bone marrow cells do not express all tissue-specific antigens (TSA) meant tolerance toward skin could not be achieved in many cases. Indeed, skin-specific alloantigens have been identified as targets for rejection.^71,72 Split tolerance toward heart allografts has also been observed in both murine and canine models of mixed chimerism.^51,55 These data suggest split tolerance is not the unique result of alloreactive T cells recognizing skin-specific antigens, and suggests other polymorphic TSAs may generate split tolerance. We recently proposed a model for split tolerance requirements, outlining the necessity of TSA polymorphisms.³⁷ Interestingly, previous non-myeloablative mixed chimeras were tolerant to islets,⁷³ seemingly suggestive of a lack of islet-specific alloantigen polymorphisms in the mouse combinations studied. However, the extent of split tolerance due to TSAs is likely to be underestimated based on existing data. Many of the model systems used to evaluate the potential for mixed chimerism-induced tolerance employ donor and recipient combinations that are matched for minor antigens, eliminating the potential for TSA polymorphism to trigger rejection of donor tissues in chimeras. Unfortunately, as mentioned before, minor antigen matching is not something that is feasible in the clinic. Even when minor antigens are not intentionally matched, it is not clear that the limited number of inbred mouse strains used in chimerism studies will reflect the extent of TSA polymorphism in the human population. It seems unlikely that TSAs in humans would lack polymorphisms. The relative rarity of split tolerance in murine studies is quite likely to be misleading. A recent study has shown that all of the various laboratory mouse strains are derived from a limited heritage and consequently have a very limited genetic diversity.⁷⁴ Therefore, it can be anticipated that inbred mouse strains greatly under-represent the true quantity of polymorphisms in TSAs that exist in wild populations. This raises the question, how many polymorphic TSAs need to be expressed in a donor tissue to generate an immune response that is sufficiently strong enough to cause rejection? Unlike the tolerance to donor bone marrow antigens (which will have the capacity to take advantage of central tolerance), tolerance to allelic variants of TSA will be fully dependent on peripheral mechanisms. Therefore, with highly immunogenic tissues such as lung and intestine and non-vascularized grafts, such as skin, a very small number of allelic tissue antigens may be sufficient. In contrast, less immunogenic organs such as kidney, heart and liver may require the donor to express many mismatched TSAs. This scenario would make split tolerance to the most commonly transplanted organs (e.g., kidney) a much rarer event. This is indeed the case in cynomolgus monkeys where kidney allografts survive long-term when given months after bone marrow transplantation.⁷⁵ We would suggest that the most plausible interpretation of this outcome is that the few allelic TSAs present in the donor kidney do transiently induce an immune response, however, the response is weak and eventually switches to a tolerogenic response as the graft heals in and APC-activating danger signals subside. Since central tolerance is not available for TSAs, peripheral tolerance must be induced, and would have to occur even after an initial state of priming to the TSAs. In this scenario, it may be that in order to avoid split tolerance toward highly immunogenic allografts (e.g., skin) the tolerizing agents (e.g., co-stimulatory blocking agents) should not only be administered during infusion of donor bone marrow, but also when the donor tissue/organ is transplanted.³⁷ It will therefore be important to not simply design the duration of our tolerance induction protocols on the length of time required to induce tolerance to the bone marrow cells, but also the length of time required to induce tolerance to TSA of the organ transplanted. Given the wide distribution within the recipient of donor antigens on bone marrow cells, and the relatively restricted location of TSA, it can be anticipated that it will take longer to induce tolerance to TSA than to bone marrow antigens. Given the existence of polymorphisms in TSA and the potential for split tolerance, it would be advantageous to develop chimerism approaches that generate at least some degree of donor-specific regulation (often referred to as ‘dominant tolerance’). In this way, regulatory T cells recognizing donor antigens common to donor hematopoietic cells and tissues can inhibit the response to TSA. This approach may involve an alteration of the class of anti-donor response in the few anti-donor T cells that manage to escape elimination, rather than a true tolerance that mimics self-tolerance.^76,77

The concept that TSAs trigger rejection is not the only possible explanation for split tolerance under all conditions. Tissue-specific alloantigens as the explanation of split tolerance for skin grafts has been questioned in the past.^78,79 Furthermore, tissue-specific alloantigens could not explain the differential susceptibility to rejection of heart, islet or skin transplants in a CD8 T cell transgenic mouse model,⁵⁶ or islet, skin and hematopoietic cell transplants in a CD4 transgenic model.¹⁷

Direct and indirect allograft rejection

Allograft recognition by recipient T cells through either the direct⁸⁰ or indirect^81,82 pathway is independently capable of transplant rejection. With high frequencies of allospecific T cells, direct recognition is thought to be the dominant mechanism of acute rejection.^83,84 However, indirect allorecognition is also relevant and is the driving force behind chronic rejection.^85,86 With indirect recognition, allogeneic MHC⁸⁴ or non-MHC^87,88 antigens can be presented in the context of host MHC to host effector cells. Furthermore, certain transplantation locations⁸⁹ and certain antigens⁹⁰ may preferentially present via the indirect pathway.

Although indirect alloresponses by CD8 T cells are possible,^91,92 CD4 T cells may preferentially respond via the indirect pathway.^93-95 Indirectly activated CD4 T cells can provide help to alloreactive cytotoxic T cells^96,97 and stimulate the production of alloantibodies by B cells.⁹⁸ In addition, CD4 T cells can act as effectors. Without CD8 T cells or B cells, CD4 T cells can mediate transplant rejection when activated solely via the indirect pathway.^17,82,87,99

The central deletion of directly alloreactive T cells in bone marrow chimeras is likely mediated by donor bone marrow derived antigen-presenting cells^5,6,100 and T cells¹⁵ that migrate to the thymus. However, mixed chimeras may not immediately become tolerant to donor antigens via the indirect pathway. Therefore, it is possible to visualize a situation where indirect immunity to transplanted tissues could potentially explain split tolerance. In an experiment to examine the role of T cells that see donor antigen in host MHC alleles (indirect immunity) in split tolerance, we created chimeras in which donor cells had both donor and host MHC alleles. In this manner, we could test if split tolerance in fully MHC mismatched chimeras might be due to the inability to directly present to and tolerize host T cells that have indirect anti-donor specificity. Even under conditions where the donor cells could directly present donor antigens to ‘indirect pathway’ T cells, the indirectly reactive T cells were tolerant of donor hematopoietic cells, but were able to reject donor skin.³⁷ This demonstrated that in mixed chimeras there is indirect tolerance toward donor hematopoietic cell, but not skin TSAs.

An additional potential mechanism of split tolerance is a differential susceptibility of allogeneic tissue types to indirect T‑cell rejection. Although there is a known hierarchy in the susceptibility of different allografts to rejection,^{56,79,101,102} it is unknown why the differences exist, and there are many possible explanations. Using a monoclonal population of CD4 T cells recognizing a defined antigen-MHC complex, indirect immunity alone is sufficient to reject skin transplants,^17,87 but not heart¹⁰³ or thymus¹⁰⁴ transplants. Furthermore, in a wild-type mouse using an MHC mismatched donor-recipient combination, orthotopic corneal allotransplants are rejected exclusively by indirectly activated CD4 T cells.⁸⁹ In a chimerism model, transgenic CD4 T cells could mount an effective indirect immune response against donor B cells, islets and skin grafts given early post inoculation with donor hematopoietic cells.¹⁷ However, within the same animal where donor B cells and skin transplants were being rejected, donor T cells were not eliminated. This demonstration of split tolerance was due to a relative resistance of T cells to undergo indirect rejection.

Direct immunity may play a role in the rejection of some tissues, however, most of the direct alloresponse should be abrogated in mixed chimerism due to central deletion. Although, with increasingly mild mixed chimerism conditioning protocols (non-myeloablative, co-stimulation blockade based) that leave more of the recipient T‑cell compartment intact, direct recognition may indeed play a larger role in allogeneic tissue rejection. Taken together, split tolerance may be generated by properties of the host (direct and indirect immunity) or the donor tissue (susceptibility to indirect recognition, and expression of TSA).

Dynamic donor phenotype

The concept that donor hematopoietic stem cells may have a phenotype that confers resistance to host immunity, either through direct contact inhibition or low expression of MHC molecules,^105-107 is receiving increased attention but remains controversial. After bone marrow transplantation for chimerism induction, donor stem cells migrate to and proliferate in the host bone marrow compartment.^3,108 These stem cells will then differentiate into myeloid and lymphoid progenitors and be exported into the periphery. Split tolerance may then be possible through a disparity of susceptibility of the differentiating stem cell progeny to rejection by the host immune system. Perhaps this is one of the contributing mechanisms of split tolerance seen in a non-obese diabetic (NOD) mouse model of chimerism.⁴⁵ In this experimental system, NOD chimeras demonstrated persistent T‑cell chimerism but rejected other donor hematopoietic cells, including B cells. This result mirrors the outcome in the CD4 transgenic T‑cell model previously discussed.¹⁷ These data demonstrate that different lineages of terminally differentiated hematopoietic cells are not equally susceptible to rejection by the immune system.

Similar to the aforementioned studies on tolerance to NIMA, cellular differentiation as a potential mechanism of split tolerance has also been seen in maternal chimerism with fetal cells.¹⁰⁹ During gestation, fetal cells may cross the placenta and establish fetal microchimerism in the mother.^110,111 However, in some instances, fetus-derived stem cells may be present at high levels in the bone marrow of sensitized mothers, but not in any peripheral organs.¹⁰⁹ This finding may be a result of the maternal immune system’s inability to recognize the ‘immune privileged’ fetal stem cells. As the fetal stem cells differentiate, the maternal immune system may then recognize the foreign cells as they upregulate MHC and reduce immunoregulatory cytokine production.¹¹² These results, in both experimentally-induced and natural chimerism, suggest that as the differentiating progeny of donor stem cells change their phenotype they can become susceptible to immune recognition and/or immune effector mechanisms. This immune recognition may then account for split tolerance toward differentiated hematopoietic cell lineages. In contrast, for more terminally differentiated tissues, a changing cellular phenotype is likely to make only a small contribution to the development of split tolerance.^110,111

Multiple mechanisms of split tolerance

The potential mechanisms behind the development of split tolerance in mixed chimerism are not necessarily mutually exclusive. In NOD mouse models of mixed chimerism, a resistance to tolerance induction^113-116 or a defect in self-tolerance^117-120 may heighten the potential for split tolerance via additional mechanisms besides those involved in split tolerance to TSAs. NOD mice require more intense conditioning to establish chimerism than other mice strains, and there is a lower incidence of lasting mixed chimerism.^{34,45,121,122} This NOD mouse tolerance resistance may decrease the efficiency of the chimerism conditioning regimen, thereby preventing complete tolerance to donor antigens.⁴⁵ In this manner, chimerism may become more dependent on peripheral tolerance, of which there are notable defects in NOD mice.^119,120

Although we hypothesize the NOD mouse strain may be more prone to split tolerance, many chimerism studies in NOD mice do not support this hypothesis. As only a few of the many chimerism protocols involved fully MHC-mismatched combinations,^123-126 partial donor-recipient MHC matching may explain the lack of split tolerance.^127,128 The difficulty in establishing mixed chimerism in NOD mice may also limit the detection of split tolerance. For example, chimerism induction can lead to initial mixed chimerism that can eventually become full chimerism, thus diminishing the probability of split tolerance.^121,125 Furthermore, split tolerance toward hematopoietic cells may be missed in cases where chimerism levels were not monitored long-term.^123,126

We developed a NOD mouse chimerism conditioning regimen with a radiation-free approach that was successful across full MHC mismatches.⁴⁵ While induction of autoimmunity in transplantation is an important potential source of split tolerance,¹²⁹ islet-specific autoimmunity could not explain the rejection of donor islets by these NOD chimeras, as syngeneic islets were not rejected. Additionally, in contrast to chimeric C57BL/6 mice, chimeric NOD mice rejected donor skin grafts even when these grafts were given on the day of bone marrow transplantation (during the tolerance-promoting co-stimulatory blockade treatment). The survival of donor T cells may be related to a resistance of donor T cells to indirect rejection,¹⁷ despite this being an unusually potent rejection pathway in the NOD mouse.^93-95

These studies identify a number of underlying factors for the development of split tolerance in NOD mixed chimeras. For example, resistance to chimerism induction, donor-recipient combinations, minimal conditioning induction protocols, tissue-specific antigens and the ability of indirect effector mechanisms to reject certain allografts could all be contributing. Our NOD mouse chimerism induction protocol across fully allogeneic barriers consistently generates split tolerance; thereby allowing us an opportunity to study the cells and mechanisms involved in the split tolerance in this model. Split tolerance is likely to be a more important obstacle to the success of chimerism approaches than previously considered, which may impact tolerance induction in islet transplantation and other types of donor tissue transplantation.

Clinical Chimerism, Transplantation and Split Tolerance

The success of hematopoietic chimerism leading to donor-specific tolerance in animal models^130-133 has translated into clinical trials. The augmentation of chimerism in the clinical setting is an attempt at decreasing the amount of immunosuppressive medications transplant patients receive, increasing graft survival and, ideally, mimicking the induced tolerance seen in mice. Although the donor organ can act as a small source of pluripotent hematopoietic cells,¹³⁴ bone marrow transplantation (BMT) is needed to provide sufficient amounts of donor hematopoietic cells to generate macro-chimerism.

Initially, piloted in only a few patients, bone marrow grafts [11x10⁹ bone marrow cells (BMC)] were given 21–25 d after kidney transplant with an immunosuppressive regime that consisted of azathioprine, anti-lymphocyte globulin and prednisone.^135,136 The early results were encouraging, with decreased levels of kidney graft rejection and decreasing levels of donor responsiveness. Subsequently, a larger study of post-kidney transplant bone marrow administration (2–3x10⁸ BMCs/kg) was done with cyclosporine, prednisone, azathioprine and anti-lymphocyte globulin immunosuppression.¹³⁷ Although not randomized, in this study there was better kidney graft survival at both 12 and 18 mo in the group that received the bone marrow transplant. However, there were no differences in renal graft function or rejection episodes between the two groups.¹³⁸ Increased long-term kidney graft survival in patients receiving BMT was also seen in a more recent series that included OKT3 induction, tacrolimus, methylprednisolone and mycophenolate mofetil.¹³⁹ Also in this study, fewer kidney rejection episodes were seen in the recipients of the BMT (7x10⁸ BMCs/kg) which was given in two doses on post-operative day four and between days 10 and 14. The success with bone marrow infusion was also seen with liver allografts. In 1997, the first randomized trial of liver allografts with peri-operative BMT showed significant results in favor of multiple donor bone marrow infusions.¹⁴⁰ Specifically, both patient and liver graft survivals were greater in patients who received multiple bone marrow infusions over controls who did not receive bone marrow or who received a bone marrow graft on the same day as the liver. In addition, this study showed that cytoablative conditioning was not necessary to improve allograft survival when the recipient was given multiple bone marrow infusions. Recently, the bone marrow after organ transplantation technique has been used as part of the immunosuppressive regime for face allografts.¹⁴¹ In another recent study, kidney transplantation under the cover of total lymphoid irradiation, antithymocyte globulin, cyclosporine, prednisone and mycophenolate mofetil followed by administration of 1x10⁶ CD3⁺ T cells and 8x10⁶ CD34⁺ enriched donor hematopoietic cells/kg was attempted.¹⁴² Impressively, this regimen established mixed chimerism and tolerance toward the allograft such that all immunosuppressive medications were discontinued six months after transplant. Subsequently, this protocol has led to stable kidney graft function and withdrawal of immunosuppressives in 8/12 patients.¹⁴³ Although bone marrow infusions may have shown a benefit for kidney, liver and facial tissue allografts, donor BMT’s were not able to increase pancreas or kidney graft survival in simultaneous kidney/pancreas transplants even though there was maintenance of peripheral blood chimerism.^144,145 The lack of effect on graft survival with simultaneous kidney/pancreas transplants is in contrast to kidney grafts alone despite similar immunosuppressive therapy (OKT3, tacrolimus, azathioprine and methylprednisolone) and bone marrow dose (5x10⁸ BMCs/kg). In the studies that monitored peripheral blood chimerism, it is evident that BMT increases the levels of chimerism over controls that do not receive BMT.^144,146 However, the split tolerance seen with the persistence of peripheral blood chimerism in the face of solid organ allograft rejection (even in the presence of immunosuppression) demonstrates complexities in the relationship between tolerance and chimerism.

In contrast to the augmentation of chimerism by BMT, donor microchimerism commonly refers to chimerism that is detectable after a solid organ graft only. In these cases, microchimerism is a consequence of the passenger leukocytes migrating out of the transplanted tissue.¹⁴⁷ As its name suggests, microchimerism is detectable at very low levels, usually only by molecular mechanisms. Although microchimerism is detectable after liver, intestinal or kidney transplant,^137,148-150 there is debate regarding its immunological consequences. The observed immunological consequences of microchimerism range from long-term donor graft acceptance and donor-specific unresponsiveness^147,151 to having no predictive value for clinical course^152-156 to being associated with graft rejection and high responsiveness to donor antigen.^157-159 As such, in humans with long-term graft acceptance and immunosuppressive withdrawal, it is not clear as to whether microchimerism is the cause or effect of operational tolerance toward donor antigens. As with macrochimerism, the split tolerance seen with microchimerism outlines the complex relationship between the presence of chimerism and tolerance. Further studies of microchimerism are required to delineate its potential use as an indicator for immunosuppressive withdrawal.¹⁶⁰

Despite the disparate clinical observations seen with microchimerism after solid organ transplant, the identification of microchimerism in long-surviving kidney and liver recipients at the University of Pittsburgh¹⁵¹ prompted a program to augment chimerism in solid organ recipients by including a peri-operative bone marrow transplant. The unique Pittsburgh method of chimerism augmentation has no patient pre-conditioning and 1–6x10⁸ BMCs are given on the same day as the organ allograft, not weeks afterwards.¹⁶¹ Using this method of bone marrow administration, heart,¹⁶² lung,¹⁶³ kidney,¹⁶⁴ pancreas,¹⁶⁵ liver and kidney/pancreas^166-168 transplants have been performed. As with other BMT along with solid organ grafts, in these studies, chimerism levels in patients that received a BMT were higher than the controls. These studies also demonstrate that there is variation in outcomes by enhancing chimerism depending on the organ transplanted. For example, some studies have shown decreases in acute rejection episodes for heart,^162,166 and an ability to decrease steroid dose in kidney, lung or pancreas recipients.^163,166,168 However, with respect to kidney transplants, delayed graft function and the incidence of acute and chronic rejection have remained similar between BMT and control patients.^161,167 Although there may be a benefit to enhancing chimerism by administering a BMT (with no pre-treatment) along with a solid organ graft, there are few attempts at immunosuppression withdrawal. Therefore, despite chimerism augmentation, true tolerance appears to be rarely achieved without a conditioning regimen.

The induction of mixed hematopoietic chimerism through non-myeloablative pre-conditioning and BMT can induce allograft tolerance in murine,⁶⁶ large animal¹⁶⁹ and non-human primate models.^75,170 However, in contrast to the mouse models, the establishment of chimerism does not guarantee solid-organ tolerance in large animal models. In addition, split tolerance may be observed much more frequently in large animals and humans than in mice; for example, in a canine model, despite high levels of chimerism, donor-specific heart transplants were rejected.⁵¹ In contrast to the loss of a solid organ, when mixed chimerism tolerizing strategies are employed in non-human primates, some kidney grafts survived long-term, despite the decline and eventual loss of peripheral chimerism.^75,170,171 The first clinical trial of non-myeloablative conditioning and BMT to induce mixed chimerism and tolerance toward kidney transplant were done in patients with multiple myeloma.^49,52 Long-term outcomes of these seven patients outline three important points.⁵⁰ First, non-myeloablative BMT can induce operational tolerance toward kidney allografts as discontinuation of immunosuppression was achieved in 4/7 patients. Second, graft vs. host disesase (GVHD) occurred in 4/7 patients, raising the importance of this complication after BMT. Third, an unusual form of split tolerance was observed, as most patients lost peripheral chimerism and yet remained tolerant of the kidney allograft. However, in one patient both renal graft tolerance and mixed chimerism was maintained. Subsequently, patient pre-conditioning and BMT for the purpose of inducing tolerance toward kidney allografts was trialed in non-malignant settings.^39,142 In this series, 4/5 patients demonstrated tolerance toward the kidney allograft after the withdrawal of immunosuppressive medications. However, mixed chimerism was short-lived with all patients losing peripheral chimerism by day 21 post-BMT, suggesting a state of split tolerance may have developed.

In contrast to experiments showing that chimerism is required for maintenance of tolerance in mice,¹⁷² the demonstrations of split tolerance in humans indicate that sustained peripheral macro-chimerism may not be necessary for long-term allograft tolerance. Perhaps, as seen with passenger leukocyte experiments¹⁷³ and pre-natal chimerism models,¹⁷⁴ induction of a certain level of hematopoietic chimerism is critical to establish rather than maintain allograft tolerance. Or perhaps, since it was not monitored,³⁹ only microchimerism is required for ongoing allograft tolerance.

Conclusions

While the strengths of chimerism-induced tolerance are well appreciated,¹ there are impediments to this approach. First, less-toxic conditioning regimens have allowed the translation of chimerism induction into the clinical setting, and importantly, into the non-malignant clinical setting. However, for the routine use of chimerism as an adjuvant for solid organ transplantation, even milder or more targeted regimens must be tried. In addition, shorter protocols must be also be tried as current conditioning regimens that involve many days of pre-treatment make chimerism a possibility only in the living donor situation. Second, the potential for development of GVHD is also a common concern with bone marrow transplantation and chimerism. Last, as seen in both animal studies and human clinical trials, the relationship between chimerism and tolerance is not straightforward. Split tolerance is a serious potential pitfall of chimerism-induced tolerance because it can affect allografts as well as donor hematopoietic cells.^{2,37,42-44,68,69} Since split tolerance affecting hematopoietic cells can lead to a loss of chimerism with unclear consequences on tolerance, further research into its mechanisms are warranted.

Glossary

Abbreviations:

BMC

bone marrow cell

BMT

bone marrow transplantation

GVHD

graft-versus-host disease

GVL

graft-versus-lymphoma

MHC

major histocompatibility complex

NIMA

non-inherited maternal antigens

NOD

non-obese diabetic

Treg

regulatory T cell

TSA

tissue-specific antigen

TBI

total body irradiation