Molecular Barriers to Xenotransplantation1

Five years ago we explored the immunologic barriers to transplanting organs between phylogenetically disparate species (1). At that time, the primary barrier to discordant xenotransplantation seemed to be hyperacute rejection (HAR*), and we concluded this process might be initiated by one or more of four factors: (1) the binding of xenoreactive natural antibodies to their corresponding antigens expressed on endothelium of donor organs; (2) the direct activation of recipient complement via the alternative pathway on donor cells; (3) the incompatibility of donor complement regulatory proteins with the recipient complement system, which would allow for unrestrained activation of complement against the newly transplanted organ; and (4) the action of host natural killer cells on foreign endothelium. Over the past 5 years many of these ideas have been tested and basic knowledge has advanced to the point where a new thesis concerning the molecular barriers to discordant xenotransplantation can be put forward. Moreover, data have emerged which demonstrate that the barrier to xenotransplantation now extends beyond HAR. Our own studies and those of other investigators indicate that HAR of xenografts can now be averted most or perhaps all of the time. When HAR is averted, the xenograft becomes subject to one or more of the biological outcomes shown in Figure 1. In this article, we discuss these potential outcomes, emphasizing our current understanding of the cellular and molecular events that occur during the xenograft rejection process. HYPERACUTE XENOGRAFT REJECTION Susceptibility to HAR of xenotransplants has been previously considered to reflect in some quantitative way the phylogenetic distance between donor and recipient species (2-6). Yet, the characteristics of HAR suggest it is a distinct entity and not simply the extreme among a range of tissue reactions that vary continuously between species. In a given combination of donor and recipient species, HAR occurs nearly always, as in guinea pig-to-rat cardiac xenografts, or almost never, as in hamster-to-rat cardiac xenografts. There seems to be no “intermediate” state of the process. Thus, susceptibility to HAR likely reflects a discontinuous or heritable characteristic. In primate recipients, it is now clear that susceptibility to HAR of porcine organs is governed by two factors: the expression of an antigen(s) by the donor which is recognized by xenoreactive antibodies, and the relative incompatibility of donor complement regulatory proteins with the recipient's complement system (7-9). These two factors have independent genetic inheritance and the expression of each trait is not strictly related to phylogenetic distance between the donor and recipient. Xenoreactive natural antibodies and the antigens they recognize. The importance of naturally occurring xenoreactive antibodies in initiating HAR is illustrated by the following observations: (1) organs undergoing HAR contain bound immunoglobulin along vascular surfaces (10); (2) depletion of xenoreactive antibodies from the primate recipient prolongs porcine xenograft survival even when the recipient's complement system remains intact (10, 11); and (3) a porcine organ transplanted into a newborn primate that lacks xenoreactive antibodies but has an intact complement system does not undergo HAR (12). These data support the conclusion that disrupting the antibody-antigen reaction helps to mitigate HAR and provides the scientific basis for efforts to identify and modify the antigens recognized by naturally occurring xenoreactive antibodies. It is now generally accepted that the pool of xenoreactive antibodies in primates that initiate the HAR of porcine organs are specific for galactose-α(1-3)-galactose (Galα(1-3)Gal). The importance of Galα(1-3)-Gal as the primary antigen is supported by the following arguments. First, antibodies from human serum that bind to porcine cell lines can be blocked by the addition of purified α-galactosyl sugars such as Galα(1-3)Gal, but not by other saccharides (13). Second, COS cells transfected with cDNA for murine α1,3-galactosyltransferase express Galα(1-3)Gal, which allows binding of human xenoreactive antibodies and complement-mediated lysis (14). Third, removal of Galα(1-3)Gal from the surface of porcine endothelial cells prevents the binding of xenoreactive antibodies and complement deposition (15). The synthesis of Galα(1-3)Gal is catalyzed by α1,3-galactosyltransferase, which is present in all lower mammals and New World monkeys but not in humans, apes, and Old World monkeys; Old World primates and man have naturally occurring antibodies specific for that structure (16, 17). Much of the early work on xenoreactive antibodies specific for Galα(1-3)Gal focused on the IgG class of immunoglobulins (18-20). However, several lines of evidence now suggest that HAR is mediated primarily by the 2-4% of human IgM specific for the Galα(1-3)Gal epitope, rather than the IgG antibodies with the same specificity. Data supporting this conclusion include the following: (1) IgM antibodies are invariably found in rejecting porcine-to-primate xenografts, whereas IgG is sometimes absent; even when IgG is present it is often found later than IgM deposition within the rejecting xenograft (10). (2) The extent to which human complement is activated on porcine endothelial cells correlates with the concentration of xenoreactive IgM and not xenoreactive IgG (21), and the activation of complement in a human serum on porcine cells is prevented by depleting IgM from the serum (22, 23). (3) Purified IgM, but not purified IgG, when applied to porcine endothelial cells, activates human complement (14, 24). (4) Infusions of purified human IgG (which include IgG specific for Galα(1-3)Gal) into primate recipients of porcine xenografts did not hasten HAR, and in fact seemed to protect the porcine xenograft from HAR (24). Susceptibility of a xenograft to heterologous complement. HAR is always associated with the activation of complement (25) and does not occur if the recipient's complement system is inhibited or if the recipient is complement deficient (26-29). In some combinations of donor and recipient species, such as guinea pig to rat (30) or rabbit to newborn pig (31), the recipient's complement system is activated spontaneously on donor cells via the alternative pathway. In these cases, HAR can occur in the absence of xenoreactive antibodies and is very aggressive, perhaps because the kinetics of complement activation is independent of antibody binding. In pig-to-primate xenografts, complement is activated predominantly via the classical pathway following the binding of xenoreactive antibodies (10, 11, 25); thus, manipulation of xenoreactive antibody binding to porcine endothelial cells has a beneficial effect on complement activation and subsequent graft survival. Regardless of how complement is activated, the xenograft is extraordinarily susceptible to complement-mediated injury. The molecular basis for this susceptibility seems to be related to the inadequate function of complement regulatory proteins, such as decay-accelerating factor (DAF) and CD59, which normally inhibit the activation of autologous complement on autologous cells, protecting the cells from inadvertent complement-mediated injury. However, in the case of pig-to-primate xenotransplantation, porcine DAF and CD59 function in a species-specific manner and are unable to down-regulate the activation of primate complement (6, 30, 32). This hypothesis was first tested in vitro when human DAF was inserted via a phosphatidylinositol anchor into porcine endothelial cells preventing cell-mediated lysis by human serum (32). These observations led to the development of transgenic animals by White (33), Fodor (34), and our research collaboration7(35), which express human complement regulatory proteins. Transplanted organs from these animals have demonstrated that even low levels of human DAF and or CD59 expression will in most cases prevent HAR in pig-to-primate xenografts7(8). These data support the conclusion that the molecular incompatibility of complement regulation is an important factor in pathogenesis of HAR and that if this molecular defect is corrected the process of HAR can be averted in most cases. The pathogenesis of HAR. HAR of porcine-to-primate xenografts is initiated by the binding of xenoreactive antibodies to donor endothelial cells followed by the activation of complement, as shown in Figure 2. Complement activation then leads to the devastating pathologic picture of HAR. In the extreme case, the development of HAR might be caused by complement-mediated lysis of endothelial cells, which, by disrupting the barrier between the components of blood and extravascular tissues, yields hemorrhage and thrombosis, the prototypic lesions of HAR. Indeed, early studies describing the dissolution of endothelium in HAR suggest that cellular lysis might be the predominant process (36). However, more recent work has suggested that lysis is not the predominant lesion during the development of HAR in porcine-to-primate xenografts (10, 25). The onset of HAR apparently depends on the assembly of terminal complement components on the endothelial cells of a newly transplanted organ, as it does not occur in C6-deficient recipients (29). HAR does occur in transgenic organs that express human CD59, even though membrane attack complex formation is notably reduced (37). Thus, terminal complexes such as C5b67 or C5b678 may be sufficient to bring about HAR. How do terminal complement complexes cause the dramatic pathologic changes of HAR in the absence of cellular lysis? We originally proposed that deposition of antibodies and sublytic amounts of complement might cause “activation” of endothelial cells, which in turn would cause the pathologic changes characteristic of xenograft rejection (1, 6). A basic understanding of the endothelial cell biology has evolved to the point where we now postulate that HAR is linked to specific cellular events that do not reflect classic changes typical of endothelial cell activation as induced by cytokines or endotoxin, although such changes may occur. Rather, HAR may represent certain novel structural and biochemical changes induced by complement. The most profound changes seen in HAR involve the loss of the barrier function of endothelium and a switch of the cell phenotype from an anticoagulant posture to one that is highly procoagulant. One noncytotoxic event that may promote these processes is an alteration in endothelial cell shape mediated by C5b67, which results in retraction of endothelial cells and formation of intercellular gaps (38). These gaps, approximately 5 μm in diameter, may allow for the escape of vascular contents and the adhesion and activation of platelets to the extracellular matrix. Attachment and activation of plate-lets in this way leads to release of vasoactive substances, such as thromboxane A2, that constrict vascular smooth muscle and alter regional blood flow. Once platelet activation occurs in the setting of endothelial cell damage or retraction, the microenvironment becomes extremely thrombogenic via the adherence and activation of membrane-binding blood-clotting factors (39). This process is exacerbated by the release of heparan sulfate from endothelial cells during the HAR process (22). Heparan sulfate proteoglycans normally enhance antithrombin III activity, which in turn inhibits blood coagulation, complement activation, and oxidant-mediated injury (40). Thus, the loss of heparan sulfate proteoglycans from the cell surface leaves the cell susceptible to procoagulant and complement-mediated injury. Furthermore, complement-mediated injury to the endothelial cell may directly induce procoagulant changes to the cell membrane, further stimulating the thrombotic process. Although this hypothesis has not been rigorously tested in the setting of xenotransplantation, Hamilton et al. (41) demonstrated that deposition of the C5b6789 complement complex results in the expression of the appropriate surface conditions for assembly of the prothrombinase complex. ACUTE VASCULAR XENOGRAFT REJECTION If HAR is averted, a xenograft becomes subject to another type of rejection, which we have called acute vascular xenograft rejection (AVR) (42). The histologic picture of AVR includes hemorrhage and thrombosis, features characteristic of HAR; however, AVR arises under conditions in which HAR is clearly avoided, such as during the treatment of the recipient with complement inhibitors or in concordant species combinations. Furthermore, the pathogenic mechanisms leading to AVR seem to be distinct from those leading to HAR. Thus, AVR is not simply a “delayed” form of HAR, as some have suggested, but rather an independent process that is likely governed by the physiologic state of the vascular endothelial cell. If HAR can be viewed as the recipient “rejecting” the endothelial cell via antibody deposition and complement activation, AVR can be viewed as the transplanted endothelial cell “rejecting” the host by the induction of procoagulant proteins and cytokines. As our understanding of the molecular and cellular events associated with AVR have matured, it has become apparent that this type of rejection is best defined as the persistent state of vascular endothelial cell activation which leads to microvascular thrombosis via the induction of procoagulant proteins, cytokines, and cell adhesion molecules. The pathogenesis of AVR. Our current understanding of the pathogenesis of AVR suggests a model where endothelial cells develop a phenotypic shift in the nature of their interaction with the blood elements circulating above. Some changes in endothelial cell function which seem to be related to AVR include: (1) a decrease in endogenous tissue plasminogen activator activity and a subsequent increase in plasminogen activator inhibitor-1 activity, resulting in a global loss of fibrinolytic activity on the endothelial cell (43); (2) down-regulation of endothelial cell thrombomodulin, resulting in a loss of anticoagulant activity via the activation of protein C (44); (3) a decrease in heparan sulfate from the endothelial surface resulting in a loss of anticoagulant activity via anti-thrombin III inhibition of blood coagulation proteases (22); (4) induction of tissue factor activity on the surface of the endothelial cells, which provides the initiating cofactor for procoagulant blood-clotting complex assembly (45-47); (5) induction of cell surface E-selectin and vascular cell adhesion molecule 1, which assist in the binding of leukocytes to the endothelial cell (48); and (6) secretion of endothelin-1 and prostacyclin, which promotes vasoconstriction and decreases blood flow within the injured vessel (49). These changes occur via the endothelial cell. In combination, they result in the pathologic lesions of vascular thrombosis and xenograft demise. Figure 3 illustrates our current understanding of the cellular response observed during AVR. Saadi et al. (47) reported recently that deposition of xenoreactive antibody and the formation of the membrane attack complex on porcine aortic endothelial cells lead to the induction of interleukin (IL)-1α and that IL-1α then seems to function in an autocrine or paracrine fashion to promote at least some of the changes in endothelial cells outlined above (47). This problem may be exacerbated by vasoconstriction and microvascular thrombosis, resulting in vascular stasis, where small amounts of endothelial cell-derived IL-1α would remain in the microenvironment creating a positive feedback loop leading to further endothelial cell stimulation. The induction of procoagulant changes on endothelial cells seen during AVR may be exacerbated by various levels of molecular incompatibility which prevent the normal down-regulation of a procoagulant environment controlled by the endothelial cells. As discussed above, the inability of porcine complement regulatory proteins to control the activation of human complement is a major problem that contributes to the process of HAR. Thus, it is likely that other porcine-derived cellular proteins may not function appropriately in down-regulating proinflammatory or procoagulant events when interfaced with human blood-derived proteins. One example of this, recently identified in our laboratory, is the function of porcine thrombomodulin with human thrombin. Under normal conditions, thrombomodulin is expressed on the surface of vascular endothelial cells and down-regulates the thrombotic process by thrombin-dependent activation of protein C (50). Activated protein C then rapidly degrades the activated procoagulant cofactors factor Va and factor VIIIa. studies the of porcine thrombomodulin to human protein as would occur in a porcine-to-primate demonstrate that porcine thrombomodulin has of the activity normally seen in the autologous system This loss of protein activity seems to be governed by the inability of human to bind to porcine thrombomodulin and the inability of human protein C to with the porcine thrombomodulin complex. The physiologic of this suggest that the microenvironment of the xenograft may be protein C deficient and this state may the graft to microvascular Although in thrombomodulin regulation represent one specific case of molecular incompatibility between man and it is likely that other molecular and the physiologic interaction of blood-derived and expressed proteins. Although within the xenograft are to represent the result of cellular changes caused by an alternative hypothesis for graft in this setting may be that deposition does not represent the result of changes induced by the vascular but rather may to promote the process. In this case, microvascular deposition would lead to within the organ, which then results in a of the tissue leading to endothelial cell and organ The of antibodies in initiating AVR. Regardless of which of events leads to the development of evidence suggests that antibodies are important in the pathogenesis of this process. Our current suggests that AVR is most likely mediated by xenoreactive antibody binding to the Galα(1-3)Gal on the porcine vascular endothelial cell. supporting this conclusion include the following: (1) levels of xenoreactive antibodies in serum increase following to porcine organs and the increase in antibody with the development of AVR and (2) when primate recipients are with that inhibit the synthesis of xenoreactive antibodies or when these antibodies are from the AVR seems to be inhibited or antibodies might the cellular changes seen in AVR by the activation of sublytic amounts of complement, leading to the release of IL-1α and a persistent state of endothelial cell activation by directly vasoconstriction and by antibody binding to specific cellular The of demonstrated that if or antibodies from the recipient of an graft survival be in of the of an intact complement system and the of or antibodies to the This was later to seemed to be an of the graft the of a of antibodies which have caused rejection of the (6, This has been to where the recipient depletion of antibodies via (10, Although the biological mechanisms that this are it is likely that this process results from a shift in the of the between the graft and the or a in the biology of the endothelial cell which becomes to injury. Regardless of the cellular mechanisms that lead to it is likely that this process will be for xenograft survival. In this would the survival of porcine endothelial cells that express the Galα(1-3)Gal in the of host antibodies against that and a normal complement system of the recipient. studies the biochemical nature of are to identify the molecular nature of this process. XENOGRAFT REJECTION of intact very evidence that the interaction of host cells with a xenograft contributes to HAR the other AVR is associated with an of host leukocytes and is some evidence that these cells may to the pathogenesis of this Indeed, recent studies have some important into the nature of the cellular a xenograft might may be susceptible to injury mediated by natural killer cells. This susceptibility reflects in the mechanisms that control cell cells express for class and under physiologic conditions the interaction of these with class inhibits the function of cells, preventing injury to autologous cells the extent that may to these it is that cells in a xenograft recipient may be from on the cells may be by antibody bound to cells. The of xenoreactive antibodies of the recipient to the graft may promote the interaction of cells with a graft How the interaction of cells with a xenograft lead to tissue the of cells for an organ might cause of endothelial and the procoagulant changes discussed above. Yet, human cells not cause to the porcine endothelial cells, but structural and on the endothelium. when human cells are to of porcine endothelial cells, they induce of the endothelial cell with formation of intercellular gaps and the development of cellular changes with the phenotype In addition to the action of cells, a xenograft may be to cell-mediated It was that the of such might be to the process of positive leading to during the of cells that cells directly might be than the of cells that might cells in this Indeed, was some evidence that direct of cells might occur at all and that cell might be to the pathway. work in several has however, that human cells can porcine cells directly Thus, the xenograft may the cellular reaction by this pathway. Furthermore, et al. recently observed that human cells that porcine class antigens the direct pathway can among antigens in such a way as to allow for How might the cell response to a xenograft from the cell response to an First, the cell response to a xenograft might be because of the nearly of and to the recipient's cells via the pathway. Indeed, the of cell response by this might the of to a xenograft a than the of to an Second, of cell adhesion and have been viewed as a that might direct and cell these might mechanisms such as and which might control cell Thus, the might shift further than previously cell-mediated Over the past 5 years our understanding of the mechanisms of xenograft rejection have As specific molecular elements that lead to xenograft rejection have been have emerged that seem to each This has led to that the of organs between discordant species is an this will be by efforts to a understanding of the molecular basis of xenograft We that efforts to the cellular to xenotransplantation will an of knowledge the mechanisms cell The biological outcomes of discordant xenograft The natural of discordant xenotransplantation leads to HAR. If HAR is prevented via of antibody binding or complement the xenograft is then subject to AVR. If AVR is the graft will likely some of At that the graft will be subject to cell-mediated rejection, rejection, or xenograft xenograft occurring antibodies specific for the Galα(1-3)Gal antigen bind to the surface of porcine endothelial cells This process is followed by the activation of complement, which leads to the formation of terminal complement complexes on the endothelial cells This then the release of heparan sulfate proteoglycans from the endothelial cell surface and small gaps allow for the binding of platelets to the and allow for the deposition of blood coagulation which leads to deposition and vascular xenograft The binding of naturally occurring antibodies on the surface of porcine endothelial cells followed by sublytic levels of complement activation attack complex may induce AVR. These events induce a phenotypic shift in the surface of the endothelial mediated in by the induction of IL-1α and leads to the induction of tissue factor the release of plasminogen activator inhibitor-1 expression of E-selectin a decrease in tissue plasminogen activator inhibitor-1 activity, and a loss in thrombomodulin activity Furthermore, endothelial cells may be directly by the binding of natural antibodies to cellular that express the Galα(1-3)Gal epitope, as illustrated by the in and