Immunologically nonspecific mechanisms of tissue destruction in the rejection of skin grafts.

When mice are lethally irradiated and reconstituted with allogeneic bone marrow cells, their skin is repopulated over a period of several months with Langerhans cells (LC) of marrow donor origin. Skin from such mice, when transplanted to unirradiated syngeneic recipients, became in many cases the sites of intense inflammatory responses that led to varying degrees of destruction of the transplanted skin and in some instances, to rejection of the entire graft. The frequency and intensity of these responses were influenced by the nature of the immunogenetic disparity between the donors and recipients of the marrow cells. Chimeric skin placed on hybrid mice derived from crosses between the marrow donors and recipients behaved in all respects as syngeneic grafts or autografts. When the recipients of the chimeric skin were presensitized to the antigens of the marrow donor, the responses were especially intense, and resulted in all cases in complete rejection. Thus the immunologically mediated attack on the allogeneic LCs was accompanied by widespread and nonspecific destruction of bystander cells. In all cases, the inflammation and tissue damage were confined sharply to the grafted skin, showing clearly that nonspecific or indirect tissue destruction is entirely consistent with highly selective destruction of grafted tissues. This finding removes a major objection to postulated mechanisms of rejection that involve indirect destruction of grafted tissues.

Xenogeneic proliferation and lymphokine production are dependent on CD4+ helper T cells and self antigen-presenting cells in the mouse.

We studied proliferation and interleukin 2 production by B6 mouse spleen cells in response to stimulation by irradiated cynomolgus monkey spleen cells and compared the results with responses against whole MHC-disparate allogeneic controls (BALB/c). We found that (a) primary xenogeneic helper responses were absent, whereas primary allogeneic responses were brisk, (b) secondary xenogeneic helper responses were dependent on CD4+ T cells and responder antigen-presenting cells (APCs), whereas allogeneic responses could be mediated by either CD4+ or CD8+ T cells independently and were primarily dependent on the presence of stimulator APCs, and (c) secondary xenogeneic helper responses were blocked by an antibody directed against responder class II MHC molecules. These results suggest that mouse helper T cells recognize disparate xenoantigens as processed peptides in association with self class II MHC molecules, similar to the recognition of nominal antigens and unlike direct allo-recognition.

Xenogeneic skin graft rejection is especially dependent on CD4+ T cells.

B6 mice were treated in vivo with anti-CD4, anti-CD8, or both anti-T cell antibodies together in an effort to prolong xenogeneic compared with allogeneic skin graft survival. Mice treated with anti-CD4 antibody showed prolonged survival of xenogeneic monkey or rabbit skin even after they had rejected whole MHC-disparate allogeneic mouse skin. Furthermore, the addition of cyclosporine was synergistic with the anti-CD4 antibody in prolonging graft survival. These results suggest that the cell-mediated response to xenogeneic antigens is especially dependent on CD4+ lymphocytes, a feature shared by the response to allogeneic minor histocompatibility antigens. In addition, the results suggest a possible approach to clinical immunosuppression for some forms of xenogeneic transplantation.

Variations in the responses of C57BL-10J and A-J mice to sheep red blood cells.

In response to repeated injections of sheep red blood cells, C57BL/10J mice produce predominantly 19S antibody in increasingly higher amounts, while A/J mice initially produce 19S antibody and then switch to produce increasing 7S antibody titers. In an F(1) generation all mice responded like the C57BL/10J mice. Backcross data implied genetic control involving at least three loci.

Lysis of antibody-coated cells by platelets.

Antibody-coated erythrocytes are lysed by murine C5- whole blood but not by plasma separated from such blood. The lytic activity has been shown to derive from platelets that attach to sensitized cells probably through membrane receptors for C3b. Whole blood or platelet-rich plasma (prp) obtained from mice that have been treated with purified cobra venom factor has little or no activity unless it is fortified with fresh C5- plasma. Lysis is observed only if the reactants are incubated at 37 degrees C and mechanical shaking is practiced, at least intermittently, throughout the period of incubation. Adherence of platelets and subsequent lysis are mediated by antibodies of a variety of immunoglobulin classes, including those that fail to mediate complement-dependent lysis. Platelet-mediated lysis is limited to cells to which the platelets adhere; 51Cr labeled, unsensitized cells that are mixed with prp and sensitized, unlabeled cells do not release 51Cr. Normal murine lymphoid cells and ascites tumor cells of mice, rats, and guinea pigs were apparently unaffected by sensitization and incubation with prp. However, because adherence of platelets to these sensitized cells was not observed, it is not clear whether the cells are resistant to the lytic action of platelets or whether the conditions of incubation were unfavorable for the attachment of platelets to the surfaces of nucleated cells. The significance of the lytic reaction described here is not known but may lie in antibody mediated release of microbicidal substances from platelets.

The vascular bed as the primary target in the destruction of skin grafts by antiserum. II. Loss of sensitivity to antiserum in long-term xenografts of skin.

Rat skin that survives for long periods of time on immunosuppressed mice becomes resistant to anti-graft serum and remains so for as long as it survives. When long-standing grafts are removed and placed on new immunosuppressed mice, they remain resistant to antiserum for as long as they survive. The acquired resistance to antiserum seems, therefore, to be due to changes in the grafts rather than to changes in their hosts. Furthermore, it was found that the acquisition of resistance is correlated with replacement of graft endothelium by host cells, as demonstrated by the use of immunofluorescent techniques in conjunction with mouse anti-rat serum and rat anti-mouse serum. Evidently, humoral antibodies are able to cause acute damage to skin grafts, and presumably to grafts to other organized tissues, only if they react with antigens of graft endothelium. Long-term grafts that are retransplanted to their original donors or to rats syngeneic with those donors are in most cases rejected, whereas 14-d-old grafts similarly regrafted are in no case rejected. Apparently, the responses of the secondary recipients to the mouse endothelial antigens in long-term grafts lead to destruction of the entire grafts. When long-standing rat skin xenografts are removed and placed on untreated mice syngeneic with the primary hosts, they are in every case rejected, although they survive slightly longer than skin taken directly from rat donors. Rejection is accompanied by a mononuclear infiltrate and is qualitatively indistinguishable from the rejection of freshly prepared rat skin. Clearly, sensitized cells are more efficient than humoral antibody in destroying grafted tissues.

Indirect recognition by helper cells can induce donor-specific cytotoxic T lymphocytes in vivo.

In vitro studies have revealed that help for cytotoxic T lymphocyte (CTL) induction can be mediated through several pathways, including direct recognition of allogenic class II antigens by CD4+ cells, direct recognition of allogeneic class I antigens by "CD4-independent" CD8+ cells, and "indirect" recognition of peptides of alloantigens presented in association with self class II molecules. Whereas good evidence for the two direct pathways is available in vivo, there is relatively little evidence to show that indirect recognition can initiate graft rejection. This study examined the role of indirect allorecognition during the generation of CTLs in mice as they rejected major histocompatibility complex (MHC) class II-deficient skin after depletion of CD8+ T cells in vivo. Recipients were depleted of CD8+ T cells by in vivo treatment with anti-CD8 monoclonal antibody and then grafted with allogeneic skin lacking MHC class II antigens. The mice rejected the skin grafts rapidly. Although flow cytometry showed marked depletion of CD8+ T cells in these mice, we found that (a) CD8+ CTLs were generated and sensitized to MHC class I antigens of the donor; (b) the generation of the CD8+ CTLs required the help in vivo of CD4+ cells, as well as priming with the allogeneic skin graft; and (c) the CD4+ T helper cells were sensitized indirectly to donor peptides presented in association with class II antigens on recipient antigen-presenting cells. These results provide evidence that indirect recognition can provide effective help for CTL induction during graft rejection, even when the cytotoxic T cells are sensitized by determinants expressed only on the donor graft.

Acute destruction by humoral antibody of rat skin grafted to mice.

A study has been made of the roles played by complement and polymorphonuclear leukocytes (PMN) in the acute destruction of xenografts of rat skin that follows injection of their hosts with antisera specifically reactive with graft antigens. The rat skin was grafted onto mice whose immune responses were suppressed by removal of the thymus and a brief course of treatment with rabbit antimouse lymphocyte serum. At about 2 wk after grafting the mice were injected intravenously or intraperitoneally with mouse antirat serum (MARS). This time interval was chosen because it avoided the complications that might be associated with either the process of healing in or with incipient rejection. Signs of graft damage were evident as early as 10 min after the injection of MARS, and in most animals so injected the grafts were completely destroyed within 24-48 h. The role of complement (C) in this acute destructive process is indicated by the results of three lines of experimentation. (a) Non-C-fixing antibodies or antibody fragments failed to cause damage to the grafts. Indeed, both chicken antirat serum and F(ab')(2) fragments from rabbit antirat serum completely protected the grafts against the effects of MARS that was administered 24 h later. (b) When mice were depleted of hemolytic C by treatment with cobra venom factor or heat-aggregated gamma globulin, the damage caused by MARS was greatly reduced or completely inhibited. (c) In mice with a genetically determined absence of C5 much greater quantities of MARS were required to cause graft damage; the tempo of the destructive process was consistently slower; and a greater number of grafts recovered from the initial inflammatory process than was the case for animals with an intact complement system. The participation of PMN in serum-mediated destruction of grafts was initially suggested by the results of microscope examination of fixed tissues. The essential role of these cells in the process is indicated by the failure of MARS to cause tissue damage in mice whose circulating PMN have been reduced to very low levels by treatment with nitrogen mustard or more specifically with an anti-PMN serum. The absence of tissue damage when circulating PMN are reduced but C levels are normal suggests that C-mediated cytolysis is unimportant in graft destruction and that the role of C lies in its ability to generate chemotactic factors. The latter may then attract the PMN that provide mediators of tissue damage.

The vascular bed as the primary target in the destruction of skin grafts by antiserum. I. Resistance of freshly placed xenografts of skin to antiserum.

Rat skin grafted onto immunosuppressed mice is resistant to mouse anti-rat serum during the first 7-10 d after transplantation. It gradually acquires susceptibility, reaching a peak of sensitivity at 14-16 d after grafting. The grafts remain sensitive to antiserum, though at decreasing levels for an additional 3 wk, and grafts that persist beyond that time are resistant to antiserum for as long as they survive. In the study reported here, it is shown that the initial period of resistance to antiserum is due to factors acting locally within the graft and is entirely uninfluenced by the regimen of immunosuppression or the protective dressings that are used. After administration of antiserum, deposits of the injected immunoglobulin and of endogenous C3 are found on the luminal surfaces of graft vessels, although no significant tissue damage is observed. Rat skin that has become highly sensitive to antiserum 14-16 d after transplantation loses that sensitivity if it is regrafted to a new recipient, and then regains it 8-10 d later. Thus, the resistance of freshly grafted skin to antisera is associated with the process of healing into place, a conclusion that is supported by the observation that the intracutaneous administration of antisera to rats causes intense local inflammation and necrosis. The skin is therefore sensitive just before it is removed for grafting, but temporarily loses sensitivity thereafter. Resistance to antiserum during the first 3 or 4 d after transplantation is probably attributable to the fact that at that time grafts are vascularized poorly if at all. The state of resistance extends for several days after vascularization of the graft takes place and is then only gradually lost, a phenomenon that seems to be associated with the resistance of newly formed and regenerating blood vessels to vasoactive substances. This view is in accord with and, indeed, supports the idea that the induction of vascular injury is an essential step in antisera-mediated damage to tissue grafts.

Fibronectin is produced by blood vessels in response to injury.

During the time of tissue repair that ensues subsequent to tissue injury, blood vessel wall fibronectin increases concomitantly with endothelial proliferation and angiogenesis. However, the source of this blood vessel fibronectin had not been delineated. In this report we have demonstrated that microvascular fibronectin is produced in situ by the proliferating vessels surrounding excisional wounds. This finding was established by extirpating 3 mm of skin from the center of a well-healed rat xenograph on the flanks of immunosuppressed mice, harvesting the injured skin sites at various stages during the healing process, and staining the specimens with reciprocal species-specific anti-fibronectin. The proliferating donor vessels that surrounded the wounded graft had increased fluorescence staining with FITC conjugated mouse anti-rat fibronectin and no staining with rat anti-mouse fibronectin. This finding was taken as direct evidence that the fibronectin was produced in situ by the rat vessels and not derived from circulating mouse plasma.

Fibronectin beneath reepithelializing epidermis in vivo: sources and significance.

Fibronectin and fibrinogen occur under the migrating epidermal tongue during reepithelialization of an excisional wound, and fibronectin increases in conjunction with capillary and fibroblast ingrowth during wound healing. Although we have previously shown that fibronectin is produced by proliferating blood vessels, the source of fibronectin associated with reepithelialization and fibroblast ingrowth has not been determined. In this report we demonstrate that subepidermal fibronectin derives mostly from plasma early in reepithelialization of an excisional wound and comes from both plasma and in situ production late in reepithelialization. This finding was established by extirpating 3 mm of skin from the center of a well-healed rat xenograph on the flanks of immunosuppressed mice, harvesting the open wound sites at 2, 4, 7, and 10 days after injury, and staining the specimens with reciprocal species-specific anti-fibronectin antibodies conjugated with fluorescein. In the first 4 days after wounding, newly forming rat epidermis migrated mainly over mouse fibronectin. In contrast, by 7 days after excision, the rat epidermis transits over a matrix containing both mouse and rat fibronectin, or rat fibronectin alone, indicating that a major component of the fibronectin is produced in situ. Although the biologic significance of these observations has not been fully elucidated, fibronectin may be part of a provisional matrix that functions to support, if not actively participate in, cell recruitment to sites of inflammation or wound healing.

Fibronectin beneath reepithelializing epidermis in vivo: sources and significance.

Fibronectin and fibrinogen occur under the migrating epidermal tongue during reepithelialization of an excisional wound, and fibronectin increases in conjunction with capillary and fibroblast ingrowth during wound healing. Although we have previously shown that fibronectin is produced by proliferating blood vessels, the source of fibronectin associated with reepithelialization and fibroblast ingrowth has not been determined. In this report we demonstrate that subepidermal fibronectin derives mostly from plasma early in reepithelialization of an cxcisional wound and comes from both plasma and in situ production late in reepithelialization. This finding was established by extirpating 3 mm of skin from the center of a well-healed rat xenograph on the flanks of immunosuppressed mice, harvesting the open wound sites at 2, 4, 7, and 10 days after injury, and staining the specimens with reciprocal species-specific anti-fibronec-tin antibodies conjugated with fluorescein. In the first 4 days after wounding, newly forming rat epidermis migrated mainly over mouse fibronectin. In contrast, by 7 days after excision, the rat epidermis transits over a matrix containing both mouse and rat fibronectin, or rat fibronectin alone, indicating that a major component of the fibronectin is produced in situ. Although the biologic significance of these observations has not been fully elucidated, fibronectin may be part of a provisional matrix that functions to support, if not actively participate in, cell recruitment to sites of inflammation or wound healing.

An isotopic antiglobulin test for quantitation of anti-sheep red blood cell antibodies of individual Ig classes in mice bearing different IgCH haplotypes.

An isotopic antiglobulin test (IAT) for anti-sheep red blood cell (SRBC) antibodies of different immunoglobulin (Ig) classes was developed and used to measure levels of IgM, IgA, IgG, IgG2a, IgG2b, and IgG3 anti-SRBC antibody in sera from 12 inbred strains of mice. Characterization of the Ig class specific heteroantisera prepared for use in this assay revealed substantial amounts of antibodies reactive with idiotypic and allotypic determinants of IgG1, IgG2a, IgG2b and IgA. Accordingly, class specific antibodies which do not discriminate among Ig molecules of different IgCH phenotypes were isolated from each heteroantiserum by absorption and elution from Ig's of selected strains. Levels of anti-SRBC antibody are expressed as units of antibody/ml relative to an anti-SRBC antiserum reference standard. The use of conversion factors (estimates of the relative reactivity of each anti-Ig reagent) allows quantitative comparisons within the between IgG subclasses. This method allows measurements of nanogram amounts of specific anti-SRBC antibodies of a single Ig class, and simultaneous measurements of antibody levels in 6 Ig classes from a single 30 microliter serum sample. Analysis of the responses of several strains of inbred mice indicates striking variations (5-31-fold) in the amounts and relative proportions of different Ig classes. The IgG subclass distributions observed appear to be associated with the IgCH but not the H-2 haplotypes of the strains studied.

Evidence that multiple defects in cell-surface molecule interactions across species differences are responsible for diminished xenogeneic T cell responses.

The purpose of the present study was to identify which of the several possible defects in cell-surface-molecule interactions are responsible for diminished mouse helper T cell responses to xenoantigens. We measured primary mouse anti-monkey, anti-pig, and anti-human proliferation in vitro in experimental systems in which potential defects were partially corrected by lymphokine supplementation and/or the use of transgenic or hybridoma cell populations. We found that the diminished mouse helper T cell responses to xenoantigens result from at least two defects in cell-surface-molecule interactions between T cells and xenogeneic APCs, specifically TCR and/or CD8 interactions with xenogeneic class I MHC molecules and accessory molecule interactions with their ligands (probably LFA-1 with ICAM-1/ICAM-2 and/or LFA-2 with LFA-3). Other investigators have identified additional defects, such as in lymphokine function across species differences. Thus, there appear to be multiple defects responsible for the diminished cellular immune response to xenoantigens.

Immunohistochemical analyses of skin graft rejection in mice. Kinetics of lymphocyte infiltration in grafts of limited immunogenetic disparity.

We have examined the nature and kinetics of T cell infiltration into murine first-set skin allografts across a variety of histocompatibility differences. Both Ly 1+2- and Ly 2+ cells migrate into grafts that differ with respect to discrete class I, class II, or minor H barriers. The lymphoid cells in the allograft did not express MEL 14 or the B cell antigen B220. The cells in the graft progress from the vascular anastomosis region into the dermis with apparent homing and clustering about hair follicles and epithelial cells at the dermoepidermal junction. Microscopic necroses appear in a patchy distribution in these areas. The intensity of the infiltrate is related to the nature of the alloantigen recognized but it does not correlate with the tempo of rejection. Class II-disparate allografts evoke a peak response (mean 62 +/- 34 T cells/0.1 mm2) that is 2-6-fold greater than class I-disparate allografts or syngeneic control peak responses (mean 14 +/- 2.4 T cells/0.1 mm2). Although a difference at a minor H antigen evoked an intense peak T cell response (192 +/- 33 T cells/0.1 mm2), the mean survival time of the allografts was significantly longer than that observed with class I or class II differences. The IL-2 receptor+ subset of activated T cells is extremely rare in syngeneic grafts, but ranges from 0 to 56% of infiltrating T cells in allografts. We conclude that all major T cell subsets contribute to the in situ response to all classes of alloantigens, but at the apparent exclusion of the MEL 14+ T cell pool and B cells.

Prevention of alloantibody formation after skin grafting without prolongation of graft survival by anti-L3T4 in vivo.

Treatment of mice in vivo with monoclonal antibodies against the L3T4 antigen (CD4 in human beings) has been shown to suppress the humoral response to several foreign antigens and to prolong the survival of allografts in some cases. Experiments were therefore performed to test whether anti-L3T4 antibody treatment would suppress alloantibody production after skin transplantation. Monoclonal anti-L3T4 antibody (GK1.5) was administered to C57BL/6 (B6) mice prior to BALB/c skin grafting. The production of B6 anti-BALB/c alloantibody was then tested after graft rejection. The results showed that: (1) graft survival of BALB/c skin on B6 mice was not substantially prolonged by anti-L3T4 treatment; (2) graft survival was significantly prolonged if mice were treated with both anti-L3T4 and anti-Lyt2 antibody; (3) the production of alloantibody following grafting was decreased by anti-L3T4 treatment and was completely eliminated if thymectomy was also performed; (4) thymectomy prolonged the effectiveness of the anti-L3T4 treatment; (5) tolerance to alloantigens presented at the time of anti-L3T4 treatment was not achieved; and (6) well-established cytotoxic antibody titers rose to higher levels after secondary grafting even with concurrent anti-L3T4 treatment, while weak antibody titers remained stable or decreased. These results indicate that L3T4+ cells are essential in providing the "help" necessary for generating humoral responses to alloantigens but that elimination of these L3T4+ cells still allows the generation of help for cell-mediated immunity. The data also suggest that class I antigens must be presented on class II molecules in order to elicit an antibody response.

Transplantation effects of a unique major histocompatibility complex class I mutation.

This study describes a novel MHC class I mouse mutant that was discovered because of loss of reactivity of its cells to monoclonal antibodies. The mutation occurred in the H-2Ks molecule and is the first in vivo mutation described that has a single altered amino acid residue (amino acid 107) distant from the regions considered to be peptide or TCR contacts. Nevertheless, skin grafts from the mutant to the parent are rejected by CD8+ T-cells. In the reciprocal direction, the mutant shows partial tolerance to parental skin grafts, suggesting that the mutant is inefficient in selecting alloreactive T-cells specific for the wild-type Ks molecule.

Coronary atherosclerosis in transplanted mouse hearts. III. Effects of recipient treatment with a monoclonal antibody to interferon-gamma.

Obstructive coronary arterial lesions in the vessels of transplanted hearts result from a complex process in which the immune response of the recipient plays a pivotal role. We have devised an experimental system in which mouse hearts, transplanted after brief treatment with mAbs to CD4 and CD8, survive and contract for many weeks. A high percentage of such hearts develop advanced, obstructive coronary lesions by 4 weeks. Migratory cells of recipient origin localize in the linings of affected vessels, and mediators of inflammation, including adhesion molecules, are present in increased amounts in characteristic locations. Histocompatibility antigen expression is also increased, and these substances may promote the formation of vascular lesions by acting as targets for immune responses. IFN-gamma synthesis has been demonstrated in grafts where it is postulated to be important in the expression of MHC molecules and macrophage activation. Here we report that continuing treatment with R4-6A2, an mAb to IFN-gamma, strikingly inhibits the formation of obstructive vascular lesions in mouse hearts transplanted to recipients incompatible for either class I or class II antigens (P < 0.0001 for the former and P < 0.03 for the latter). Immunohistologic studies showed reduction of the class II-positive mononuclear infiltrate, but focally enhanced endothelial class I expression remained. The mechanism for this effect of anti-IFN-gamma probably extends beyond the influence of anti-IFN-gamma on increased expression of histocompatibility antigens.

Monoclonal antibody therapy. Anti-idiotypic and non-anti-idiotypic antibodies to OKT3 arising despite intense immunosuppression.

The frequency, timing, and specificity of the humoral antibody response to a murine monoclonal antibody (OKT3, IgG2a) were measured in 21 consecutive renal allograft recipients. These patients received i.v. OKT3, 1-5 mg/day for 10-20 days as treatment for acute graft rejection. Maintenance immunosuppression consisted of azathioprine and corticosteroids. Using three different assays, an antibody response was detected in 75% of the 20 patients with adequate samples. The ELISA assay of the overall IgM and IgG reactivity to OKT3 revealed that IgM anti-OKT3 appeared in 65% and IgG anti-OKT3 in 50% of the patients, reaching a peak 20-33 days after the last dose of OKT3. The IgM preceeded the IgG in most cases (P less than 0.02) and in 8 cases was detected during therapy. One patient had high levels of IgM anti-OKT3 before therapy, yet responded normally to OKT3. Interference with the therapeutic effectiveness was evident in one patient who developed IgG antibodies during therapy. His serum blocked the binding of F-OKT3 to normal lymphocytes in the presence of normal BALB/c serum. The blocking assay, done by flow cytometry, measured anti-idiotypic (Id) reactivity since the sera did not affect the binding of OKT8 (another IgG2a) or anti-Leu4 (another anti-T3), and the blocking activity remained after affinity absorption with normal mouse IgG. Using this assay, 60% of the patients made an anti-Id response. One made only anti-Id, and several had anti-Id at times when other reactivities were undetectable. Antibodies to non-idiotypic, presumably isotypic, determinants represented on OKT8 occurred in only 44%, while other reactivity (OKT4; IgG2bK) was less common (12%) and weaker. While no adverse allergic reactions occurred in this group of patients, the anti-Id antibodies, which are a prominent feature of the immune response to this and probably other monoclonal antibodies, can block their therapeutic effectiveness and can arise despite intense immunosuppression. This response may require the use of different idiotypes for prolonged or repeated courses of therapy and may be the major obstacle to the use of human monoclonal antibodies.

Xenograft rejection of class I-expressing transgenic skin is CD4-dependent and CD8-independent.

B10.PD1 mice are transgenic animals expressing a class I MHC antigen of pigs. B10.PD1 skin graft survival on B6 mice was prolonged by anti-CD4 antibody treatment in vivo but not by anti-CD8 treatment. These results suggest: (1) that the defect in cell-mediated recognition of xenografts involves an interaction with the xeno-MHC antigens themselves and not with the cells which express them; (2) that antigen processing and presentation of xeno-antigens on responder-type APCs is required in vivo; and (3) that CD8-independent, possibly noncytotoxic, mechanisms of xenograft rejection may exist.