Accommodation: Preventing Injury in Transplantation and Disease¹

In the mid-1980s, a few groups of physicians undertook renal transplantation across ABO barriers (1, 2, 3). Transplanting organs carrying blood group A and/or B Ags into recipients producing Abs against those Ags was a bold endeavor, because experience had shown that ABO-incompatible renal transplants would likely fail because of severe, unremitting rejection (4, 5, 6). To limit the humoral reaction against the ABO-incompatible kidneys, the recipients of these renal transplants were transiently depleted of anti-blood group Abs by various means (7). The results of these experiments were remarkably gratifying. In several series, the ABO-incompatible renal transplants survived and functioned as well as ABO-matched transplants (1, 2, 7).

Why renal transplants in recipients transiently depleted of Abs against the donor should succeed was not immediately clear. Several of the recipients with normal functioning grafts had anti-blood group Ab in the blood (7) and target tissues containing bound Ab (8). The Ag was not abolished nor the endothelium replaced, because the transplants contained the foreign blood group Ag expressed at high levels on the endothelium of blood vessels (8). Even more striking perhaps, a similar phenomenon was observed in a few porcine organs transplanted into nonhuman primates (9). This suggested that successful engraftment was not due to tolerance or loss of the antigenic target but rather reflected some other biological condition. We called this condition, in which an organ transplant appears to flourish in the face of humoral immunity directed against it, “accommodation” and speculated this condition could be important not only for overcoming humoral barriers to allotransplantation but also the more daunting barriers to xenotransplantation (9). In this paper, we review the mechanisms that might allow a graft to survive in the face of an immune response directed against it and consider the possibility that accommodation could have broader implications for understanding the physiology of the immune system.

Broadly construed, accommodation reflects a biological change or a coordinated series of changes that enable cells and tissues to persist in the face of noxious agonists, cells, or other factors directed against them. As a mechanism of successful engraftment, accommodation must be distinguished from tolerance, in which the immune response to a graft or foreign Ag is selectively abrogated resulting in generalized nonresponsiveness.

Accommodation may have been mistaken at first for tolerance. In 1955, Woodruff and Simpson (10) found that a second skin allograft placed on a recipient carrying a primary long-term surviving skin allograft from the same donor strain was sometimes rejected even though the primary graft was maintained. This observation led to the conclusion that the recipient tolerated the first graft but not the second. However, a more plausible explanation might focus on the possibility that the graft acquired resistance to injury by the alloimmune response, implying that the graft was accommodated.

As a working definition of accommodation, one generally expects to demonstrate the existence of an immune response against a graft in which accommodation is thought to exist and in some circumstances the immune response actually induces accommodation. Still, from a theoretical perspective, we should not want to exclude the possibility that facets of accommodation may occur in the absence of an immune response.

Beyond the impact on transplanted organs, accommodation could be vital for host defense and failure of tumor surveillance. For example, one might envision that accommodation would allow a vigorous cytotoxic response (i.e., cytolytic T cells specific for viral peptides) to control viral replication or maintain latency without causing irreversible damage to infected cells. Conversely, the fulminant destruction of organs and tissues in some conditions could reflect a failure of accommodation. Consistent with this concept, the genesis of accommodation could allow tumors to escape control by the immune system and thus to grow and spread.

Accommodation could originate from a single mechanism or from multiple mechanisms encompassing changes to both the graft and the recipient. For convenience, we shall consider accommodation of the graft to refer to changes originating in and impacting on the transplanted organ or tissue. We shall consider accommodation of the host to refer to changes in the immune system of the recipient, other than tolerance. This view is undoubtedly a simplistic one, because a graft might induce changes in the immune response just as the immune response might induce protective changes in the graft.

One way, in principle, to distinguish changes in the graft from changes in the host is to transfer a graft in which accommodation is thought to exist to a second, cyclosporine-treated recipient (preventing cellular rejection) and to transplant a fresh graft into the recipient from which the accommodated graft was taken (Fig. 1). If the accommodated graft survives in a new host while a fresh graft is destroyed in the original recipient, one would conclude accommodation reflects changes in the graft. In contrast, if a fresh graft survives in the host of an accommodated graft (in the presence of a demonstrable host immune response against the graft) and the accommodated graft is destroyed by the secondary recipient, then one would conclude that accommodation reflects changes to the immune response of the primary graft recipient. In practice, this model is exceedingly difficult to apply, and one cannot exclude the possibility that both forms of accommodation are necessary, at least to some extent, for prolonged graft survival.

Accommodation might arise if Ags in a graft were lost or were to change in such a way that Ab binding decreases. Studying guinea pigs that have the Forssman Ag on endothelium, Yuzawa et al. (11) found that administration of small amounts of anti-Forssman Ab causes modulation or shedding of the Forssman Ag. In this model, guinea pigs primed with increasing amounts of anti-Forssman Ab shed the Forssman Ag and survive administration of 20–40 lethal doses of Ab. The structure of an Ag might also change following transplantation. Carbohydrate Ags might be expressed on different core structures that present the saccharide in such a way that Ab binding is diminished (12, 13, 14). However, binding of IgM and IgG to accommodated allografts and xenografts is similar to grafts undergoing humoral rejection (15, 16, 17), and therefore, a dramatic change in Ag is probably not a general mechanism of accommodation.

Another mechanism of accommodation of the graft might involve heightened control of complement by increased removal of terminal complement complexes or inhibition of the complement cascade. Campbell and Morgan (18, 19) found that the influx of Ca²⁺ into cells following the insertion of the complement component C9 in plasma membranes initiates vesiculation of the plasma membrane containing membrane attack complexes. Through this mechanism, and perhaps through other changes, cells exposed to Abs and subtoxic amounts of complement acquire resistance to cellular lysis by a second application of complement at cytolytic concentrations (20, 21, 22). Shibata et al. (23) found that endothelial cells exposed to sublytic complement increase expression of decay accelerating factor (CD55), an inhibitor of the C3 convertase. As another example, Dalmasso et al. (24, 25) found that endothelial cells treated with IgM specific for Ags on the cells or with a lectin that recognizes Galα1,3-Gal, the saccharide targeted by Abs causing xenograft rejection, acquire resistance to complement-mediated injury. Resistance in this system was thought to be due, in part, to heightened expression of CD59, an inhibitor of the membrane attack complex (24, 26).

Accommodation of the graft might occur through regeneration of naturally protective substances such as heparan sulfate (27, 28). Under physiologic conditions, heparan sulfate imparts a negative charge to the cell surface that inhibits complement activation (27, 28, 29). Heparan sulfate also prevents extravasation of plasma proteins, injury by oxidants, and thrombosis, all of which contribute to the picture of vascular rejection. We have found that freshly transplanted grafts rapidly shed heparan sulfate (Fig. 2). If temporary depletion of components of the humoral immune system were to allow regeneration of heparan sulfate on cell surfaces and extracellular matrices, this regeneration could, in part, account for accommodation. Consistent with this possibility, we have found heightened expression of heparan sulfate in accommodated grafts.³

Another mechanism of accommodation in a graft is through acquired resistance to injury mediated by complement or other factors underlying acute vascular rejection. Fig. 3 offers a model of three potential means by which such protection might be manifest. The action of a noxious agonist or Ab on endothelium or other cells in the graft might cause the receptors for those agonists to be desensitized. As one example, Mizel et al. (30) found that ligation of the IL-1R on T cells and fibroblasts causes the receptor to be internalized, thus desensitizing the cells to further stimulation. Another mechanism involves the regulation of inflammatory or coagulant pathways. For example, exposure of endothelial cells to TNF-α induces IκB, which inhibits NFκB activation (31). Consistent with this possibility, Saadi et al. (32) found that complement activation on endothelial cells also induces IκB expression. The third mechanism involves the up-regulation of protective proteins and pathways in response to a noxious agonist providing cells or tissues of a graft with a general resistance to the initial agonist as well as other unrelated stimuli.

The importance of protective proteins and pathways leading to general resistance to injury in accommodation was presaged by Nath et al. (33) who found that rats treated with subtoxic doses of hemoglobin, an inducer of heme oxygenase-1, resist lethal kidney damage caused by glycerol. Vogt et al. (34) showed that exposure to nephrotoxic serum protects rats from subsequent glycerol-induced acute renal failure also thought to be due to heme oxygenase-1 expression. The protective mechanisms described in these models of renal injury were extended to accommodation by Bach et al. (35), who found that accommodated grafts also express heme oxygenase-1 in addition to the antiapoptotic genes A20, Bcl-2, and Bcl-x_L, whereas grafts undergoing acute vascular rejection express the proapoptotic genes Bad and Bax.

Heme oxygenase-1 acts as a powerful antioxidant, anti-inflammatory, and antiproliferative molecule by catalyzing production of other cytoprotective substances such as carbon monoxide and biliverdin (36). The importance of heme oxygenase-1 in accommodation has been shown by Soares et al. (37) who found that they could induce accommodation in wild-type but not heme oxygenase-1-deficient cardiac xenografts. The rejection of cardiac xenografts deficient in heme oxygenase-1 hints at its importance in accommodation; however, it is possible that heme oxygenase-1 may be required for survival of a graft in the face of even limited immune injury. Consistent with this possibility, the only living patient reported with heme oxygenase-1 deficiency was found to have severe endothelial cell damage independent of any immune-mediated injury (38). Soares et al. (37) did attempt to address this problem by grafting heme oxygenase-1-deficient hearts into allogeneic Rag2^−/− mice; however, this does not adequately recapitulate the conditions in which a graft undergoes accommodation.

Although these studies make a compelling argument that heme oxygenase-1 and possibly other protective proteins are needed for graft survival in the face of ischemia or other insults, they do not prove that these proteins alone mediate accommodation. Consistent with this possibility, Park et al. (39) found no increase in heme oxygenase-1, Bcl-2, or Bcl-x_L expression by microarray analysis or immunohistochemistry in long-term surviving renal allografts and accommodated ABO-incompatible long-term surviving renal allografts in human recipients. Tabata et al. (40) found heightened expression of Bcl-2, Bcl-x_L, and heme oxygenase-1 in a lung xenograft model 2 days after transplantation, but expression was found to decline to near baseline levels 21 days later, suggesting that sustained expression of these protective genes may not be necessary for maintenance of accommodation. Unfortunately, efforts to induce accommodation by expressing protective proteins have not been met with success (and because negative results are difficult to publish, these failures remain as anecdotes).

Accommodation in some cases may reflect changes in the host—particularly qualitative changes in the immune response that mitigate the injury that an immune response might cause. One change that could lessen the injury caused by humoral immunity is a shift in IgG subclass to IgG2 in humans or IgG2b in mice. IgG2 activates complement inefficiently and may block binding of other, more cytotoxic subclasses. For example, human IgG2 binds to carbohydrate Ags blocking activation of complement by IgM and IgG1 directed against the same Ag (41). Mohiuddin et al. (42) found that mice bearing accommodated cardiac grafts produce noncytotoxic Abs against the graft, whereas mice without accommodation, and thus rejecting a graft, produce cytotoxic Abs against the graft, suggesting that this mechanism could underlie accommodation.

Another change in the host that could account for accommodation is a shift in the cytokine milieu. Bach et al. (35) reported that accommodation of hamster hearts transplanted into rats was associated with a shift from a Th1 to Th2 response, because accommodated grafts contain recipient Th2 T cells secreting IL-4, IL-10, and IL-13. The authors asserted that these cytokines might confer protection, by inducing expression of heme oxygenase-1, Bcl-2, and Bcl-x_L. Subsequent studies by Bach and colleagues (43) found a gradual increase in the numbers of Th2 T cells infiltrating accommodated xenografts over time which correlated with the development of accommodation. How a Th2 response is generated in an accommodated graft is unclear, but it has been suggested that the graft might initiate the response. Delikouras et al. (44) showed that immortalized porcine endothelial cells exposed to human IgG produce high levels of NO. Dorling et al. (45) subsequently showed that NO produced by endothelial cells can induce a Th2 cytokine response in CD4⁺ T cells. Although the presence of a Th2-skewed cell infiltrate in accommodated grafts may promote survival, others have suggested that Th1 cytokines are also important. Wang et al. (46) found that rat cardiac xenografts in mice lacking IL-12 or IFN-γ, prototypical Th1 cytokines, undergo acute vascular rejection more quickly than cardiac xenografts in wild-type mice.

Another mechanism in which changes in the host could induce accommodation relates to the presence or absence of T cell help. CD4⁺ T cells recognize peptides presented by MHC class II molecules on B cells and promote the production of high affinity Ab and changes in Ab effector functions through CD40-CD40 ligand interactions (47). In the absence of T cell help, the magnitude and affinity of the Ab response may be reduced, prolonging graft survival. For example, Tanemura et al. (48) showed that mice deficient in Galα1,3-Gal, a major xenoantigen, mount a strong Ab response against the carbohydrate Ag when immunized with porcine kidney membranes but not when immunized with allogeneic murine kidney membranes expressing the carbohydrate Ag. The authors suggest that allogeneic kidney membranes contain fewer peptides that can be recognized by Th cells and thus provide insufficient T cell help for the induction of a vigorous Ab response. This suggests that accommodation might be more easily induced in ABO- and/or MHC-incompatible grafts compared with xenografts due to differences in T cell help that favor the former. Support for this hypothesis stems from the difficulty encountered when trying to induce accommodation in presensitized xenograft recipients. Presensitized rats receiving hamster cardiac xenografts and mounting a strong IgG Ab response require splenectomy, cyclophosphamide, complement depletion, cyclosporine, and blood exchange to induce accommodation (49). In contrast, successful accommodation has been observed in presensitized human renal allograft recipients using more limited induction protocols (50).

Accommodation may require a host response that gives rise to complement activation; thus, blocking or depleting Abs that activate complement may prevent accommodation. McKane et al. (51) were unable to induce resistance to complement-mediated injury, a process we believe to be necessary for accommodation, in porcine endothelial cells using anti-Galα1,3-Gal Abs of the IgG2 isotype. These results suggest that the induction of accommodation may require complement activation by IgG1 or IgM Abs. Further supporting the need for complement activation, total depletion of IgM Abs increases the difficulty of inducing accommodation in some models. For example, porcine endothelial cells preincubated with heat-inactivated human sera as a source of xenoreactive natural Abs resist complement-mediated injury upon re-exposure to Ab and complement; however, depletion of IgM Abs from the preincubation sera prevents the induction of resistance (25). These results suggest that a delicate balance between the production of complement-fixing and non-complement-fixing Ab may be required to prevent acute vascular rejection while inducing accommodation.

Accommodation of the host might also be seen if peripheral tolerance were to develop. Peripheral tolerance could manifest itself as decreased or absent production of Ab directed against donor Ags as a result of B cell tolerance. For instance, Chong et al. (52) and Lin et al. (53, 54) showed that treatment of xenograft recipients with cyclosporine and leflunomide, a regimen reported to induce accommodation, prevented the initial T-independent Ab response which was replaced with a T-dependent Ab response that appeared to exert less specificity for the graft. Yin et al. (55) subsequently showed that third-party grafts of a different species placed after cessation of leflunomide treatment were rejected via T-independent Ab responses, whereas xenografts from the same species as the primary graft survived, suggesting that a change in the recipient’s immune response had ensued. Chong et al. (52) correctly differentiated this condition from graft accommodation and thus termed it “host accommodation,” a condition akin to B cell tolerance.

We have considered accommodation as a biological change that protects a graft from destruction; however, accommodation may have a biological cost (56). We have speculated that if the accommodated state were purely protective, then it would probably be constitutive (56). That it must be induced suggests that accommodation may be detrimental. There are at least three ways the accommodated state may be detrimental to the host. First, protective genes can produce injurious substances. For example, the degradation of heme by heme oxygenase-1 yields free iron and carbon monoxide (57). Free iron functions as a powerful oxidant leading to cellular injury if it is not quickly bound by ferritin (56). Carbon monoxide dilates blood vessels, which may protect from ischemia but also generates oxidants that may cause irreversible DNA damage.

Excess accommodation might prevent the immune system from efficiently clearing pathogens. Thus, cells with heightened resistance to cytotoxicity could allow viruses to persist in the face of an immune response that would normally eliminate the virus. Racanelli and Rehermann (58) postulate that failure of CTLs to eliminate hepatitis C virus in some cases may be due to resistance of infected hepatocytes to cytotoxic effector mechanisms. Consistent with this possibility, He et al. (59) found large numbers of CD8⁺ T cells specific for hepatitis C viral peptides in the livers of chronically infected patients. These results suggest that failure to completely clear the virus from the liver is not due to a failure to mount a robust immune response against the virus but may reflect resistance of hepatocytes to the antiviral functions of CTLs.

Accommodation may also prevent the clearance of tumors as mentioned above. Reiter et al. (60) found that activation of complement in sublytic amounts on erythroleukemia cells protects the cells from subsequent attack by complement as well as perforin and melittin. Junnikkala et al. (61, 62) found that ovarian tumors and glioblastomas secrete large amounts of factor H and factor H-like protein 1 in response to treatment with Ab and serum and resist lysis by complement. Donin et al. (63) found that human ovarian, breast, and prostate carcinoma cells resist complement-mediated lysis following exposure to sublytic complement due, in part, to increased expression of sialic acid, CD59, and a protein kinase C-dependent pathway. These changes that have been observed in accommodated organ grafts could stand in the way of immune surveillance as conceived by Burnet (64).

We have used the term accommodation to refer to absence of injury in the face of humoral immunity directed against an organ graft; however, the mechanism(s) and concept of accommodation may extend much further. We would contend that accommodation likely reflects a broader process in which the tissues and organs in which immunity is manifest change in ways that inures tissues and organs to injury as a byproduct of the response. Inducing accommodation might thus allow the immune system to eliminate infectious organisms while preserving autologous cells and tissues. We hypothesize that early facets of immune responses, such as the acute inflammatory reaction or complement activation, may induce accommodation, which in turn confers resistance to subsequent, more lethal immune reactions. As an example, the self-reactive Abs produced by B1 B cells might activate sublytic amounts of complement and in this way precondition a tissue to the more cytotoxic, high affinity Abs that will follow.

Understanding how accommodation arises and functions might allow application for the treatment of disease. For example, if Abs or pharmacotherapy can be used to induce accommodation, then it might be used to treat chronic vascular diseases. For instance, Hancock et al. (65) used cobalt protoporphyrin to induce a state similar to accommodation in a mouse cardiac allograft model of chronic rejection and transplant arteriopathy. In this model, Ig-deficient mice injected with alloreactive Ab develop transplant arteriosclerosis and reject their grafts, whereas preconditioning of recipients with cobalt protoporphyrin before injection with alloreactive Ab prolongs graft survival. Injury of endothelial cells by humoral immunity and complement activation might underlie atherosclerosis (66, 67). Protective mechanisms similar to those seen in accommodation might be used to prevent the development of atherosclerosis.

Immunologists have understandably focused on how immune reactions are controlled from the perspective of cells of the immune system; however, a complete understanding of this control must also include how cells targeted by immunity and bystander cells resist inadvertent injury. This understanding will not only provide a clearer picture of the physiology of the immune system, but also teach us how this resistance, i.e., accommodation, can be directed at diseased tissues or how accommodation of tumors could be abolished.