The Immunologic Response to Xenografts

David H. Sachs, M.D.. is the Paul S. RussellAVamer-Lambert Professor of Surgery at Massachusetts General Hospital and the director of the Transplantation Biology Research Center, Harvard Medical School. Boston. The rejection of xenografts is clearly an immunologic phenomenon, since in the absence of an immune response, animals accept transplants even across widely disparate xenogeneic barriers. For example, nude mice, in which the absence of the thymus leads to defective T-cell immunity, have been shown to accept xenogeneic skin grafts ( Manning and others 1973 ), even to the point of growing chicken feathers. Similarly, some of the lymphohematopoietic compartments in severe combined immune-deficient (scid) mice can be repopulated by highly disparate xenogeneic hematopoietic cells ( Mosier and others 1988 ; McCune and others 1988 ). In theory, eliminating the immune response to a xenograft should be sufficient to assure its success. This premise is of more than just theoretical importance to the field of xenotransplantation, since there certainly might have been other, non-immunologic barriers that could have prevented xenotransplantation even in the absence of an immune response. For example, the cell surfaces of xenogeneic tissues might have been physiologically incompatible, or the red cells of the recipient might have been physiologically incapable of delivering oxygen to xenogeneic tissues or even unable to negotiate xenogeneic capillaries. Nevertheless, at least between mammalian species, it now seems likely that the immune response is the barrier of greatest importance. As is the case for other immune responses, the reaction to xenografts involves both humoral and cellular immunity. Xenografts have been further categorized as concordant or discordant on the basis of phylogenetic distance and vigor of the immune response ( Caine 1970 ). The most notable immunologic difference between concordant and discordant xenografts involves the presence in the latter of natural antibodies capable of causing hyperacute rejection of vascularized organs. As will be described in more detail below, although natural antibodies have posed a formidable barrier to discordant xenografting in the past, there are now numerous methods for eliminating these antibodies or controlling their effects, which have shown promising results in avoiding hyperacute rejection. However, both humoral and cellular immune responses to xenografts will undoubtedly be as strong as, or stronger than, responses to allografts, and will have to be overcome if xenotransplantation is to become a reality. I will review here our present understanding of these two arms of the immune response for both concordant and discordant xenografts and will try to identify strategies that may overcome the resultant barriers that each of these responses poses to successful xenotransplantation. The most widely studied concordant xenograft systems involve closely related rodent species such as the mouse and rat. In many respects, the humoral response to transplants between such species is similar to that observed for MHC-mismatched allotransplants, and as implied by the definition of concordant species, hyperacute rejection is not observed when primarily vascularized transplants are performed. However, the absence of natural antibodies between concordant species is relative rather than absolute, and even in the rat-mouse system, natural antibodies have been detected when carefully sought ( Aksentijevich and others 1991a ). Thus, using flow cytometry and cytotoxicity assays, it has been determined that normal mouse serum contains natural antibodies with specific binding to, and cytotoxicity against, scid rat bone-marrow cells ( Aksentijevich and others 1991a ). These natural antibodies were predominantly of the lgM and IgG3 classes, and activity toward bone marrow cells was much greater than that toward spleen cells. Such antibodies probably explain the observation that much greater numbers of rat than of murine bone-marrow cells are required to achieve engraftment in mice ( Ildstad and Sachs 1984 ). To more directly evaluate the effect of these natural antibodies on engraftment of rat bone-marrow cells in mice, adoptive transfer studies were performed using T- and B-cell-deficient scid mice as recipients ( Aksentijevich and others 1991b ). Because of their immunodeficiency, scid mice accepted rat bone-marrow cells readily, with only a low dose of whole body irradiation being necessary for conditioning. Passive transfer studies showed that normal mouse serum could markedly inhibit the engraftment of rat bone-marrow cells even in this phylogenetically close species combination, consistent with the hypothesis that natural antibodies provide a barrier to engraftment of xenogeneic bone marrow. Natural antibodies are probably present in other concordant species combinations, including primates. This should hardly be surprising, given the existence of ABO antibodies even within the human species. For the most part, these antibodies are probably of sufficiently low titer that they do not lead to hyperacute rejection. Nevertheless, pretransplant cross matching will certainly be required in order to avoid those situations in which relatively high titers of such antibodies might lead to catastrophic loss of a transplanted organ. As has been demonstrated for ABO antibodies, it should be possible to eliminate hyperacute rejection due to such natural antibodies by plasmapheresis or absorption techniques ( Alexandre and others 1987 ). In contrast to natural antibodies, the induced humoral response across concordant species barriers is likely to pose a major barrier to xenotransplantation. In the case of rat anti-mouse responses, immunization similar to that used to produce alloantisera was effective in raising high titers of xenoantibodies very rapidly ( Sachs and others 1971 ). Within such xenoantisera considerable titers of antibodies reactive with mouse MHC alloantigens were detected, along with other antibodies of equal or greater titer reacting with species-specific antigens ( Sachs and others 1971 ). The initial humoral response following immunization consisted predominantly of IgM, and this shifted to IgG after 2-3 weeks, suggesting a typical T-cell-dependent response ( Davie and Paul 1974 ). Similarly, humoral antibodies have been a major feature of the response to concordant xenografts in several other systems. For example, the rejection of heterotopically transplanted hearts from cynomolgus monkeys to baboons was correlated with the development of cytotoxic antibodies in the recipients' sera ( Sadeghi and others 1987 ). For this reason, splenectomy and cyclophosphamide treatment or both have been used to prolong xenograft survival in several concordant species combinations by diminishing antibody responses, presumably by removing or inhibiting antibody-producing B-cell populations ( Edwards and Rose 1989 ; Thomas and others 1992 ). Antibody was likewise a prominent feature of the immune response of human-to-baboon liver transplants, which was apparently controlled by high-dose immunosuppressive medications ( Starzl and others 1993b ). It therefore seems clear that the measures that will have to be taken to avoid the induced antibody response to concordant xenografis will have to be as least as vigorous as they are for MHC-mismatched allografts. One fortunate note is that because the predominant problem following sensitization appears to be the T-cell-dependent IgG response, it seems possible that induction of tolerance at the T-cell level (see below) will carry with it tolerance at the B-cell level with respect to this induced humoral response. The cellular response to concordant xenografts is, in general, similar to that of allografts ( Auchincloss 1988 ). In vitro cellular assays such as the mixed lymphocyte reaction (MLR) and the cell-mediated lympholysis (CML) assays have been shown to be qualitatively and quantitatively similar for concordant combinations of primate ( Martinis and Bach 1977 ), canine ( Hammer 1991 ) and rodent ( Ildstad and others 1984 ) species to the corresponding assays within each species. Since the acute rejection of untreated vascularized allografts is predominantly a cell-mediated phenomenon, as one might expect, survival times for vascularized concordant xenografis are similar to those for MHC-mismatched allografts ( Auchincloss 1988 ). Likewise, immunosuppressive agents such as anti-thymocyte globulin and Cyclosporin A, which are successful in controlling allogeneic cellular rejection, have also been effective in suppressing the cellular response to concordant xenografts ( Russell and Monaco 1967 ; Sadeghi and others 1987 ). Indeed, as early as 1964, Reemtsma and colleagues reported survival of a chimpanzee kidney in a patient for 9 months using azathioprine, Actinomycin C, steroids, and irradiation of the kidney transplant (all of these being the recommended treatment for allotransplants in that era) as the only immunosuppression ( Reemtsma and others 1964 ). The long-term survival obtained suggests that for the chimpanzee-to-human combination, current drug therapy for preventing allotransplant rejection would probably suffice. However, since chimpanzees are now considered an endangered species, it is highly unlikely that these animals will see further use as clinical concordant xenograft donors. The two clinical baboon-to-human liver xenografts reported recently from Pittsburgh were likewise treated with a four-drug immunosuppressive regimen including three agents previously used in varying combinations for allotransplantation (FK506, prednisone, and prostaglandin E). Cyclophosphamide was added to the treatment for its potential effect on the humoral response. Although neither patient survived long-term (70 and 26 days, respectively) it was noteworthy that both showed little or no evidence of cellular rejection in biopsy or autopsy specimens ( Starzl and others 1993a , 1994 ). Infectious complications were the immediate cause of death in both patients, and the investigators attributed this complication at least in part to over-immunosuppression. It therefore remains unclear whether or not levels of immunosuppression identical to those that are effective for allotransplantation would have sufficed to avoid cellular xenografi rejection for this concordant baboon-to-human xenograft combination. As noted above, one of the most important differences between concordant and discordant xenografts is that discordant xenografts contain preformed or natural antibodies capable of causing hyperacute rejection of vascularized organs. In general, the further the phylogenetic distance between two species, the greater the detectable levels of such preformed antibodies ( Hammer and others 1973 ), which increases the likelihood of hyperacute rejection. Although natural antibodies are defined as those that are present in the absence of immunization, it is likely that these antibodies actually represent cross-reactions between antibodies directed against bacterial cell-wall antigens and antigens on the surface of xenogeneic cells. Consistent with this hypothesis is the fact that germ-free animals have been shown to have very low levels of natural antibodies ( Hammer 1987 ). The majority of natural antibodies are of the IgM subclass ( Hammer 1987 ; Latinne and others 1994 ). It has been clear for some time that these antibodies are cross-reactive and bind to carbohydrate determinants on glycoproteins and glycolipids ( Galili and others 1984 ), and there is increasing evidence that the majority of natural antibodies in the human anti-pig combination are directed to the cd-3-galactose linkage ( Oriol and others 1993 ; Sandrin and others 1993 ). IgM antibodies are known to be excellent activators of complement, and the effects of these antibodies on xenografts often involve activation of the complement system ( Dalmasso and others 1991 ). For this reason, several approaches have been taken to avoid hyperacute rejection by eliminating the complement-mediated cytotoxicity pathway. These include use of complement inhibitors ( Pruitt and others 1991 ; Miyagawa and others 1993 ) and potentially the production of transgenic pigs bearing species-specific complement inhibitory molecules such as decay accelerating factor ( Langford and others 1994 ). However, natural antibodies also appear to be effective in activating endothelial cells, which may play a major role in the hyperacute rejection process ( Platt and others 1991 ), and which may not be avoided by complement inhibition. The fact that natural antibodies are predominantly IgM may be advantageous from the point of view of xenotransplantation, since IgM responses are generally primary and do not involve long-term immunologic memory. Memory for antibody formation is generally thought to occur at the stage of the IgM to IgG switch and to involve T-cell help ( Davie and Paul 1974 ). Therefore, one might hope that if natural antibodies can be removed prior to xenotransplantation, and if T-cell tolerance to xenografts can then be induced, natural antibodies may not recur. As indicated in Table 1 , there are several procedures which have been suggested as potential ways to eliminate xenoreactive antibodies. Alexandre and colleagues have pioneered the use of extensive plasmapheresis as a means of removing natural antibodies ( Alexandre and others 1989 ). In our own laboratory we have chosen an absorption technique using a pig liver to absorb antibodies in vivo prior to the xenotransplant ( Latinne and others 1993 ; Tanaka and others in press ). A one-hour perfusion was found sufficient to remove the vast majority of natural antibodies and to eliminate hyperacute rejection. More recently, we have begun to use columns in which an insoluble matrix bearing otl-3-gal epitopes is substituted for the liver in our perfusion step, and the results are very encouraging (Sablinski and others unpublished data). Strategies for eliminating natural antibodies Plasmapheresis Absorption Organ perfusion Insolubilized antigen Natural Synthetic Anti-lg Class-specific (e.g. anti-lgM) Anti-ldiotype Anti B kcell/Plasma-cell Rx Induction of B-cell tolerance Plasmapheresis Absorption Organ perfusion Insolubilized antigen Natural Synthetic Anti-lg Class-specific (e.g. anti-lgM) Anti-ldiotype Anti B kcell/Plasma-cell Rx Induction of B-cell tolerance Strategies for eliminating natural antibodies Plasmapheresis Absorption Organ perfusion Insolubilized antigen Natural Synthetic Anti-lg Class-specific (e.g. anti-lgM) Anti-ldiotype Anti B kcell/Plasma-cell Rx Induction of B-cell tolerance Plasmapheresis Absorption Organ perfusion Insolubilized antigen Natural Synthetic Anti-lg Class-specific (e.g. anti-lgM) Anti-ldiotype Anti B kcell/Plasma-cell Rx Induction of B-cell tolerance The induced antibody response to discordant xenografts may pose a more difficult problem for xenotransplantation than do natural antibodies. The induced antibodies that occur in response to such transplants are predominantly IgG and result from a T-cell-dependent response. As such, they involve long-term memory, so that absorption procedures are unlikely to provide a lasting solution to their elimination. It is not yet clear how much of the induced response is directed to the same determinants that are detected by natural antibodies and how much is directed to additional antigens. Clearly, immunization across disparate species barriers leads to antibodies to many surface molecules, with antigenic determinants carried both by proteins and carbohydrates. Methods aimed at avoiding T-cell responses to xenografts should also mitigate against the generation of induced antibodies. Perhaps the most effective way to eliminate this problem will be to avoid it, by inducing tolerance at the T-cell level, as discussed below. The primary cellular response to all transplants takes time, since both an afferent response (that is, sensitization and proliferation) and an efferent arm (mobilization of effector T cells) are required. Since natural antibodies cause hyperacute rejection within hours, most in vivo studies of discordant xenografts have concentrated on humoral and complement-mediated mechanisms rather than on the cellular immune response. However, in vitro cellular responses to xenogeneic cells have been studied extensively. Early studies produced the surprising result that cellular immunity to xenogeneic antigens appeared to be weaker than the corresponding responses to allogeneic antigens ( Wilson and Fox 1971 ; Engelhard and others 1988 ; Widmer and Bach 1972 ; Simonsen 1967 ). Such analyses suggested lower precursor frequencies both for proliferative and for cellular cytotoxic responses. On the basis of these lower frequencies, it was hypothesized by Jerne (1971) that the primary reactivity of T cells in the immune response was directed toward alloantigens (that is, minor variants of self-MHC antigens). However, there are numerous other reasons why cellular reactivity across discordant barriers could appear weaker than alloreactivity besides a difference in the T-cell receptor repertoire. One other potentially significant factor is the species-specificity of some accessory molecule interactions. For example, it is now clear that CD4 molecules on T-helper cells add to the affinity of the interaction of these cells with stimulator cells expressing class II antigens by interacting directly with the constant portion of class II molecules ( Gay and others 1988 ). CD4 molecules are highly conserved within species, but divergent between species ( Pames 1989 ), and this is also true of the constant portions of class II molecules ( Klein ). Therefore, on the species combination the interaction between CD4 and class II antigens could lead to frequencies of cells, because only T cells bearing with high affinity for xenogeneic class II to even a CD4 class II interaction would be as Similarly, one might as which are for activation ( and others 1992 ), to on the of the interaction across the species difference possible for lower for xenogeneic may be the species-specificity of ( and others 1991 ). studies by Auchincloss and colleagues using the mouse as the species and and pig skin grafts as the discordant transplants, have demonstrated that many of these potential may explain the rejection response ( and others 1992 ). These have demonstrated that the of CD4 class II and class I in these species combinations leads to use of the for sensitization to xenografts ( and others ), that is, of xenogeneic class I on class molecules ( and others 1994 ). this were true for all xenogeneic cell-mediated responses, one might actually the xenograft reaction to be to than an which both and of On the other with the of skin grafts on mice, the cellular responses that have been for discordant xenografts in vivo have been and stronger than those for allografts ( Auchincloss 1991 ). In more studies in several including our have indicated that the human anti-pig cellular response is both by and of ( and others 1994 ; and others 1993 ; and others in press ). Therefore, at least in this highly discordant species combination, it is likely that at least as as those required to rejection will undoubtedly be the of the discordant xenograft response, and the fact that even for allografts the of immunosuppressive the transplant patient on the between rejection and the of immunosuppression that will be required to avoid xenograft rejection may be so that many would to For this reason, it seems likely that the of clinical xenografting will at least in part, on ways of inducing tolerance across xenogeneic barriers rather than on immunosuppressive In our we are the use of mixed as a means of inducing tolerance across xenogeneic barriers. The we have is on in allogeneic and concordant xenogeneic systems in ( Ildstad and Sachs 1984 ; and Sachs 1989 ; and others ). The has also recently been to an allogeneic system in ( and others in press ). In T cells are from the recipient and sufficient is to for bone-marrow cells to In contrast to the use of bone marrow as a treatment of in which case of bone-marrow is such is neither necessary when bone marrow is used as a it is advantageous to achieve a of mixed in which the presence of specific antigen cells normal Such mixed long-term specific tolerance to transplants from the In our initial studies using mixed to and concordant xenograft we used irradiation and with of and bone-marrow cells ( Ildstad and Sachs 1984 ). These studies demonstrated that mixed specific tolerance to other However, irradiation is a regimen to be considered for clinical have therefore more recently to a regimen that mixed and tolerance the use of irradiation ( and Sachs 1989 ; and others ). have demonstrated that treatment of recipient mice with antibodies to the two T-cell CD4 and by irradiation and a dose of irradiation to the thymus engraftment of a of allogeneic bone and the production of mixed allogeneic ( and Sachs 1989 ). Such animals long-term mixed in all lymphohematopoietic compartments and are by from mixed by irradiation and with of T-cell allogeneic bone marrow. have been shown to be mixed and to be as demonstrated by long-term of skin grafts and rejection of a very similar we have now demonstrated mixed and long-term tolerance to kidney allografts in a cynomolgus ( and others in press ). are now to use a similar to tolerance across the discordant xenogeneic barrier of As a we are using which were in this laboratory a as both a for studies of ( Sachs 1992 ) and a potential xenograft ( Sachs 1994 ). The being is in 1 have recently reported the results of our studies using this ( Latinne and others 1993 ; Tanaka and others in press ). cynomolgus monkeys were treated with anti-thymocyte globulin to remove T-cell and cells. were treated with similar to that used in our rat-mouse and bone marrow from the pig the time of the was from the a pig liver and to the recipient for one using this a pig xenograft was transplanted as a for the induction of transplant results to have shown no of hyperacute rejection by the and kidney grafts have survived as as However, we not achieve of mixed xenogeneic in this was long-term tolerance studies have pig and in order to engraftment of pig bone-marrow These studies at the time of this but are since the recipients have normal for more than with only a pig xenograft Because xenografts are accepted by mice that cellular immune it seems likely that successful even across discordant species will predominantly on effective of immune responses. However, species-specificity of and accessory molecule may also have to be considered for long-term success. immune responses to xenografts many to immune responses that have been studied for allografts, they also some differences for discordant species barriers. to for discordant xenografts has concentrated on the of humoral immunity. However, cellular responses in vivo are at least as strong or stronger than their allogeneic For this reason, the of immunosuppression required to avoid xenograft rejection is likely to lead to an high of It therefore seems likely that the of clinical xenografting will at least in part, on ways of inducing tolerance across xenogeneic barriers rather than on immunosuppressive One such that is being in this laboratory involves the use of mixed lymphohematopoietic to specific and the so are early but