The receptor for advanced glycation endproducts (RAGE), a multiligand member of the immunoglobulin superfamily, interacts with proinflammatory AGEs, the products of nonenzymatic glycation and oxidation of proteins; high-mobility group box 1 (HMGB1), also known as amphoterin and S100/calgranulins to amplify inflammation and tissue injury. Previous studies showed that blockade of RAGE suppressed recruitment of proinflammatory mechanisms in murine models. We tested the hypothesis that RAGE contributes to alloimmune responses and report that in vivo, acute rejection of fully allogeneic cardiac allografts in a murine model of heterotopic cardiac transplantation is significantly delayed by pharmacological antagonism of RAGE. In parallel, allogeneic T-cell proliferation in the mixed lymphocyte reaction is, at least in part, RAGE-dependent. These data provide the first insights into key roles for RAGE in allorecognition responses and suggest that antagonism of this receptor may exert beneficial effects in allogeneic organ transplantation.
The receptor for advanced glycation endproducts (RAGE) is a member of the immunoglobulin superfamily and is expressed by cells linked integrally to the inflammatory response (1). Key roles for ligand-RAGE interaction in mediating migration and activation of monocytes/macrophages have been illustrated (2). Recently, we reported that RAGE was expressed by CD4+ and CD8+ T lymphocytes, and B lymphocytes (3,4). Integral to immune responses, dendritic cells (DC), endothelial and other vascular cells express RAGE (5). RAGE was first described as a receptor for advanced glycation endproducts, the products of nonenzymatic glycation and oxidation of proteins, such as carboxy(methyl)lysine adducts which accumulate in diabetes, renal failure and aging (1,6). Subsequent studies revealed that RAGE was a signal transduction receptor for distinct proinflammatory ligands that characteristically accumulate in inflammatory disorders, including high-mobility group box 1 (HMGB1) or amphoterin (7,8), S100/calgranulins (9) and Mac-1 (10). Consistent with key roles for ligand-RAGE interaction in mediating inflammatory mechanisms, blockade of RAGE suppressed inflammation in murine models of delayed-type hypersensitivity (9), bovine type II collagen-induced arthritis (11), experimental autoimmune encephalomyelitis (EAE) (3), autoimmune diabetes in NOD/scid mice (4), peritonitis (12) and massive liver injury (5). These studies highlighted important roles for RAGE in modulation of the immune response triggered by exogenous agents or by autoimmune mechanisms.
Recent observations suggested roles for the ligand-RAGE axis in DC properties and function. First, it was demonstrated that RAGE was expressed predominantly in DC in the massively injured liver (5). Further, the expression of the RAGE ligand S100 was demonstrated in DC in inflamed human kidney allografts (13). DC also express the RAGE ligand, HMGB1. In vitro, release of HMGB1 by DC modulates clonal expansion, survival and functional polarization of naive T cells, at least in part through activation of MAP kinases and NF-κB (14).
These findings suggested that RAGE played key roles in inflammatory mechanisms and led us to test the hypothesis that RAGE modulated alloimmune responses. We report for the first time that antagonism of the ligand-RAGE axis significantly delays the time to acute rejection in a murine model of fully MHC-mismatched vascularized heart grafts. In vitro studies employing murine T cells and human peripheral blood mononuclear cells (PBMC) revealed that blockade of RAGE suppressed proliferation in mixed lymphocyte reactions (MLRs).
Materials and Methods
Male C57BL/6 mice (The Jackson Laboratory, Bar Harbor, ME) and male CD-1 mice (Charles River Laboratories, Wilmington, MA), 8–12 weeks of age, were used in all experiments and maintained in a temperature-controlled room with alternating 12-h light/dark cycles. All experiments were approved by the Institutional Animal Care and Use Committee of Columbia University. These studies conform to the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health.
Antibodies and proteins
Monoclonal antibodies to human CD3 (UCHT-1), CD28 (clone CD28.2) and isotype control mouse IgG1,κ (anti-TNP) (clone 107.3) were purchased NA/LE (no azide/low endotoxin) from BD Pharmingen (San Diego, CA). The monoclonal anti-RAGE antibody 1B1H11 was generated in our laboratory (4).
Preparation and administration of soluble RAGE
Murine soluble (s) RAGE is the extracellular domain of RAGE (15) that binds RAGE ligands and acts as a ligand decoy (15). Murine sRAGE was prepared in a baculovirus expression system and purified to homogeneity by chromatography onto fast pressure liquid chromatography (16). Soluble RAGE was devoid of any contaminating endotoxin by serial chromatography onto Detoxi-gel columns (Pierce, Arlington Heights, IL) and sterile-filtered (0.2 μm) before administration. Mice were treated with the indicated concentration of sRAGE or vehicle, equal volumes of phosphate buffered saline (PBS), by intraperitoneal administration, beginning one day prior to surgery until sacrifice. Murine sRAGE was diluted in PBS. In all cases, mice received a total volume of 100 μL/day of the indicated treatment.
Mixed lymphocyte culture assays
Murine MLR: Mice were sacrificed and spleen homogenates were prepared and cells were passed through a 70-μm cell strainer (BD Falcon, NJ). Responder spleen T cells from B6 mice, enriched for T cells with a Dynal T-cell negative isolation kit (Dynal Biotech, WI) (>90% CD3+ by flow cytometry; BD Pharmingen), were cultured in triplicates in 96-well microtiter plates to assess proliferation in the presence or absence of sRAGE or anti-RAGE IgG (1 × 105 responder T cells and 0.5 × 105 irradiated (20 Gy) MHC II+ cells) in 250 μL of culture medium per well. MHC II+ cells were obtained with anti-MHC class II microbeads (Miltenyi Biotec Gmbh, Germany) from splenocytes of CD-1 mice for MLR and of B6 mice for syngeneic control cultures. As determined by flow cytometry, >90% staining for I-A/I-E (BD Pharmingen) was seen following this enrichment procedure. Proliferation was assessed as described below.
Human MLR: Human PBMC were isolated from whole blood that was anticoagulated with edetic acid by Ficoll density centrifugation. PBMC (1 × 105) were added to 96-well flat bottomed plates in Roswell Park Memorial Institute medium (200 μL). For mixed lymphocyte culture assays, cells were incubated with irradiated (25 Gy) allogeneic PBMC and stimulated with soluble monoclonal antibodies to CD3 ± anti CD28 IgG [1 μg/mL] in the presence of (s) RAGE or anti-RAGE IgG. Cells were cultured at 37°C in a humidified incubator and after 5 days pulsed for 16 h with 3H-thymidine (3.7 × 104 Bq/well). All tests were done in triplicates. Cells were recovered and measured in a liquid scintillation counter. Results are reported as stimulation index (SI; proliferation above background) ± standard error of the mean (SEM) of the triplicates.
Heterotopic, allogeneic cardiac transplantation
Abdominal heterotopic cardiac transplantation was performed in mice as previously described (17). Briefly, the donor ascending aorta was sutured end-to-side to the recipient abdominal aorta and the donor pulmonary artery was anastomosed to the recipient inferior vena cava. Transplanted hearts were assessed daily by abdominal palpation. Cessation of heartbeat was defined as the clinical endpoint and confirmed by laparotomy.
Immunohistochemistry and immunofluorescence
Sections, 5 μm in thickness, were prepared from paraformaldehyde-fixed and paraffin-embedded transplanted hearts. For paraffin-embedded tissue sections, microwave antigen recovery was performed in citrate buffer (10 mM; pH 6.0 for 25 min). Rejection was confirmed histologically on day 3 after transplantation by routine histology (hematoxylin and eosin staining). Day 3 after transplantation was chosen for histopathology as the extensive necrosis in the hearts of animals on day 7 that received vehicle treatment rendered it difficult to discern specific antigens by immunohistochemistry. Area of inflammatory cell infiltration and edema was measured on one section per mouse (n = 7 for vehicle and n = 9 for sRAGE treatment) by immunomorphometric analysis on a Zeiss Axioskop with KS300 3.0 Software (Carl Zeiss Microimaging, Inc., Thornwood, NY). Apoptosis was assessed by immunostaining for cleaved caspase 3 (Asp175) (Cell Signaling Technology, Beverly, MA). Apoptotic cell death in the graft was quantified by counting cleaved caspase 3-expressing cells on 10 high-power fields (×400) per section on four different transplants per group. The number of apoptotic cells per mm2 was reported. Lymphocyte infiltration into the graft was visualized with polyclonal rabbit anti-human CD3 antibodies (A0452, DAKO, Copenhagen, Denmark, Europe) and counted on 10 high-power fields (×400) per section on four different transplants per group. The number of CD3+ cells per mm2 was reported. Immunostaining was performed using the biotin-avidin peroxidase kit from Vector (Burlingame, CA) according to the manufacturer's recommendations. Target antigens were visualized using 3, 3′-diaminobenzidine (Sigma, St. Louis, MO) and sections were lightly counterstained with hematoxylin. Immunofluorescent staining for RAGE was performed using antimurine RAGE antibody (purified rat monoclonal IgG2A, clone 175410, R&D systems, Minneapolis, MN) and S100 immunofluorescence was achieved with anti-S100 antibody (rabbit polyclonal antibody, IgG fraction of antiserum; Sigma). Affinity-purified anti-HMGB1 polyclonal antibodies were purchased from BD Biosciences Pharmingen. Cells expressing RAGE, S100 and HMGB1 were counted on 10 high-power fields (×400) per section on four different transplants per group for HMGB1. For detection of RAGE and S100 epitopes, three vehicle-treated transplants and four sRAGE-treated transplants were examined. The number of positive cells per mm2 was reported. Biotinylated anti-mouse and anti-rabbit antibodies from Vectastain ABC kits (Vector Laboratories, Inc., Burlingame, CA) and Streptavidin-Alexaflour555 Conjugate from Molecular Probes (Eugene, OR) were used. We employed Vectashield Mounting Medium with DAPI to counterstain DNA. A Zeiss Axioskop 2 Plus microscope and Axiovision 4.4 software were used to analyze fluorescence signals. Ischemic heart transplants were excluded from study analysis by the pathologist (MS) on the basis of histopathologic findings as in myocardial infarction, such as the presence of neutrophils and necrosis and the absence of lymphocytic infiltrates.
To determine whether sRAGE administration to heart graft recipients was associated with survival, we utilized the Kaplan-Meier Product Limit Estimate and a log rank test to discern differences among groups. For comparing sRAGE at various doses with IgG repeated measure analysis of variance techniques were utilized. Further multiple comparisons were generated by Mixed Modeling methodology in SAS PROC MIXED. For immunohistochemistry and pathological analysis of allograft tissues, the Student's t-test was utilized to compare the average number of counts per mm2 between sRAGE- and vehicle-treated mice. p Values were not adjusted for multiplicity of testing. For all statistical analyses, data were analyzed using the SAS system software (SAS Institute, Inc., Cary, NC).
Antagonism of the ligand-RAGE axis modulates rejection of allogeneic cardiac grafts in a murine model
To test the hypothesis that RAGE contributed to alloimmune recognition, we employed an established model of heterotopic allogeneic heart transplantation. Fully mismatched grafts (donor, CD1 strain, H2q) were transplanted into C57BL/6 recipients (H2b). Mice were treated with sRAGE, 100 or 200 μg/day, beginning one day prior to transplantation and continued once daily until sacrifice. Control animals received equal volumes of PBS. No other immunosuppressive or immune-modulating agents were administered during the study. Cessation of heartbeat was defined as the clinical endpoint. The mean graft survival time in vehicle (PBS)-treated mice was 7.3 ± 0.7 days (Figure 1). In contrast, mice treated with sRAGE, 100 μg/day, displayed a significant increase in graft survival, 11.7 ± 1.7 days. Furthermore, mice treated with the higher dose of sRAGE, 200 μg/day, demonstrated an even greater graft survival time of 19.5 ± 2.8 days; p < 0.001 (Figure 1).
Ligand-RAGE expression in allogeneic cardiac grafts: Impact of soluble RAGE
The significant effect of sRAGE on graft survival led us to assess the impact of RAGE antagonism on the expression of RAGE proinflammatory ligands and the receptor itself. Immunofluorescence microscopy was employed to assess levels of RAGE, S100 and HMGB1 on day 3 after allotransplantation. An 83% reduction in cells expressing RAGE was observed in sRAGE-treated graft recipients. There was a mean number of 121.5 ± 16.7 RAGE-staining cells per mm2 in the vehicle-treated hosts, versus 20.1 ± 4.9 in hosts that received RAGE blockade; p < 0.0001 (Figure 2A–C, respectively). As previous studies linked RAGE's ligands to proinflammatory states, we assessed levels of S100 and HMGB1 in cardiac allografts. S100 staining was significantly reduced by 62% in hosts that received sRAGE. The mean number of S100 expressing cells per mm2 in graft sections of vehicle-injected mice was 99 ± 7.7 compared to 33.3 ± 2.4 in the animals receiving the soluble receptor; p < 0.0001 (Figure 3A–C, respectively). In addition, we examined expression of HMGB1 in the grafts. Compared to vehicle-treated grafts on day 3, in which the mean number of HMGB1-expressing cells was 13.6 ± 5.2, in sRAGE-treated animals, the mean number of HMGB1-expressing cells was reduced by 58%, to 5.7 ± 3.0; p = 0.0405 (Figure 4A–C, respectively).
Blockade of the ligand-RAGE axis suppresses inflammatory cell infiltration and apoptosis in the allograft
The significant increase in graft survival time in sRAGE-treated allograft host mice led us to examine the impact of RAGE blockade on immune cell infiltration into the allografts. We performed histological analysis on day 3 after transplantation in each group of mice. Compared to PBS-treated animals, mice treated with sRAGE, 100 μg/day, displayed significantly less edema and inflammatory cell infiltration by H&E staining (Figure 5A,B, respectively). Multiple sections were prepared from distinct mice and subjected to quantification of areas of inflammatory cell infiltration and edema in the allograft by morphometric analysis. Compared to PBS-treated mice, in which mean ratio (%) of inflamed myocardium versus total myocardial area examined was 43.2 ± 5.4%, animals treated with sRAGE, 100 μg/day, displayed a highly significant 79% decrease in rejection-related inflammation, 8.9 ± 10.1%; p < 0.0001 (Figure 5C).
Consistent with the significant decrease in rejection, we observed that compared to PBS treatment, in which the mean number of cleaved caspase-3-expressing cells in the allograft was 21.6 ± 6.4, mean numbers per mm2 were significantly reduced in sRAGE-treated allografts, 9.1 ± 3.1; p = 0.0127 (Figure 6–C, respectively).
As these data pointed to key roles for RAGE in modulating alloimmune reactions, we tested the impact of sRAGE on T-cell infiltration into the allografts. We observed a significant decrease in T-cell infiltration, as assessed by immunostaining with anti-CD3 IgG in sRAGE-treated mice versus those animals receiving PBS (Figure 7A–C). The mean number of CD3+ cells per mm2 graft section in the vehicle-treated group was 129.6 ± 14.8 compared to 11.6 ± 1.7 in the sRAGE receiving hosts (p = 0.0005).
MLR: Modulation by the ligand-RAGE axis in murine T cells
The finding that significantly less T cells were present in the sRAGE-treated allografts led us to test the hypothesis that RAGE directly modulated alloimmune responses. To investigate the role of the RAGE-ligand axis on donor-reactive T-cell priming, we employed purified T cells and MHC class II+ antigen presenting cells retrieved from MHC-mismatched mice. We employed distinct strategies to probe the role of the RAGE axis. We tested sRAGE and blocking antibodies to the receptor. Incubation with sRAGE, 90 μg/mL, resulted in a statistically significant 93% decrease in lymphocyte proliferation versus IgG control-treated cultures; p = 0.01 (Figure 8A). The effects of sRAGE were dose-dependent, as incubation with lower doses of sRAGE, 30 and 10 μg/mL, resulted in no significant change in lymphocyte proliferation versus nonimmune IgG; 52% and 2% decrease; p = 0.08 and p = 0.85, respectively (Figure 8A).
To substantiate the specific role of RAGE, we incubated cells in the mouse allogeneic mixed lymphocyte culture with blocking antibodies to RAGE or nonimmune IgG. Compared to nonimmune IgG, incubation with monoclonal anti-RAGE IgG, 90 or 30 μg/mL, resulted in a statistically significant 98% and 71% decrease in lymphocyte proliferation; p = 0.01 and p = 0.02, respectively. In contrast, in the presence of a lower concentration of anti-RAGE IgG, 10 μg/mL, no significant effect on mouse T-cell proliferation was noted; 13% decrease; p = 0.22 (Figure 8B).
These findings indicated that sRAGE suppressed donor reactive T-cell priming responses.
MLR: Modulation by the ligand-RAGE axis in human PBMC
Finally, based on the impact of RAGE blockade on suppression of MLRs in murine T cells, we sought to extend these findings and test the potential effect of RAGE antagonism on human lymphocyte proliferation triggered by alloresponses. Incubation with sRAGE, 100, 50 or 10 μg/mL, resulted in a statistically significant 62%, 51% and 31% decrease in lymphocyte proliferation versus control-treated cultures; p < 0.0001, p < 0.0001, p = 0.0004, respectively (Figure 8C). We next assessed the impact of RAGE blockade in cells stimulated with the allogeneic mixed lymphocyte culture in the presence of antibodies to CD3. Compared to treatment with nonimmune IgG (50 μg/mL), treatment with sRAGE, 50 or 5 μg/mL, resulted in a significant 63% and 35% decrease in lymphocyte proliferation; p < 0.0001 and p = 0.0007, respectively. In contrast, incubation with a lower dose of sRAGE, 1 μg/mL, resulted in no significant change in lymphocyte proliferation versus nonimmune IgG; p = 0.1407 (Figure 8D). Furthermore, when cells were stimulated with antibodies to both CD3 and CD28, an 82% and 25% decrease in cell proliferation versus nonimmune IgG-treated cells was noted in the presence of sRAGE, 50 and 5 μg/mL; p < 0.0001 and p = 0.005, respectively. In contrast, incubation with sRAGE, 0.5 μg/mL, had no significant effect on lymphocyte proliferation; p = 0.1207 (Figure 8E).
To directly test the role of RAGE, we incubated cells in the allogeneic mixed lymphocyte culture with blocking antibodies to RAGE or nonimmune IgG. Compared to nonimmune IgG, incubation with anti-RAGE IgG, 50 μg/mL, resulted in a 63% decrease in lymphocyte proliferation; p = 0.0002. In contrast, in the presence of lower concentrations of anti-RAGE IgG, 5 or 0.5 μg/mL, no significant effect on human lymphocyte proliferation was noted; p = 0.4827 and p = 0.8954, respectively (Figure 8F).
These findings indicated that RAGE contributed to lymphocyte proliferation in the alloresponse. The impact of RAGE did not depend on the stimulation of cells with anti-CD3 IgG alone, or in the additional presence of anti-CD28 IgG.
Our findings represent the first demonstration of roles for RAGE in alloimmune responses. In previous studies, such as delayed-type hypersensitivity, collagen-induced arthritis, EAE, peritonitis and type 1 diabetes, a significant impact of RAGE blockade on migration of inflammatory cells into inflamed foci and cytokine generation was established. In those studies, roles for RAGE signaling were implicated in both macrophages and T cells (3–5,9,11,12). The present work extends those concepts to alloimmune-mediated inflammation and suggests potentially important roles for RAGE in antigen presentation and/or processing mechanisms. Furthermore, although earlier studies suggested that T-cell RAGE might exert potent effects on inflammation and spinal cord dysfunction in EAE, and in the development of type 1 diabetes in NOD/scid mice, those studies did not dissect if the impact of RAGE blockade was primarily secondary to its effect on cellular migration and/or on activation. Here, in this ex vivo setting, triggered by alloimmune stimulation in MLR reactions, such as that induced by donor-reactive T-cell priming, these data suggest direct roles for RAGE in T-cell activation and proliferation.
Our data reveal that both S100/calgranulins and HMGB1 epitopes were highly expressed in the vehicle-treated allografts and suppressed by the ligand-binding decoy, sRAGE. S100/calgranulins are primarily intracellular molecules which may be released into the extracellular milieu in states of cellular activation. Specifically, S100/calgranulins have been found to be highly expressed in such diverse immune/inflammatory disorders as Kawasaki disease, psoriasis, inflammatory bowel disease, cystic fibrosis, arthritis and glomerulonephritis (18–23). The present experiments extend this work to alloimmune graft rejection in a murine model. Of note, other receptors for S100/calgranulins distinct from RAGE may play functional roles in certain settings. For example, although extensive data in immune and vascular cells suggest that RAGE is the chief receptor for at least certain members of the S100/calgranulin family; other studies have suggested that there may be distinct receptors for these molecules. For example, it was reported that the effect of S100b on myotube properties appeared to be RAGE-independent, as evidenced by studies using blocking reagents to RAGE and introduction of cytoplasmic tail-deleted mutant dominant negative RAGE (24).
Our findings highlight increased expression of HMGB1, particularly in a cytoplasmic location, in alloimmune settings in vivo. Like S100/calgranulins, HMGB1 is primarily an intracellular molecule. In contrast to S100/calgranulins, HMGB1 is primarily located in the nucleus, where it functions as a nonhistone DNA-binding protein. Studies have indicated that release of HMGB1 by activated cells, such as macrophages, may contribute integrally to inflammation and injury (25). In vivo, blocking antibodies to HMGB1 significantly improved survival of mice with massive sepsis (25). Since the first report of those studies, HMGB1 has now been implicated in distinct inflammatory settings as enterocolitis, lung inflammation, ischemia/reperfusion (I/R) injury in the liver and arthritis (26–29). Studies have established that in macrophages, RAGE is a chief signaling receptor for HMGB1 (30). The identification that HMGB1 may interact with toll receptors 2 and 4 suggests that the biology of this ligand may be complex (31). Nevertheless, studies presented here in the human and murine MLR experiments clearly indicated that blocking antibodies to RAGE attenuated proliferation stimulated by the allostimulus. In previous experiments in delayed-type hypersensitivity induced by methylated bovine serum albumin (BSA), in addition to sRAGE, blocking F(ab′)2 fragments of anti-RAGE IgG also exerted potent anti-inflammatory effects in the mice sensitized and challenged with methylated BSA (9).
Although not directly addressed by these studies, it is likely that RAGE-Mac-1 interactions contributed to the extensive inflammatory cell influx into the vehicle-treated allografts (10). Previous experiments indicated that sRAGE bound Mac-1, thus, it is likely that Mac-1 expressing cells were targets for the beneficial impact of sRAGE demonstrated in the present experiments. Interestingly, in published studies, the effects of RAGE-Mac-1 on migration were amplified when cells were incubated with S100b as well; these experiments suggest that the ligands of the receptor may exert additive or synergistic effects when present (10). Our data contribute to the growing body of evidence that immune/inflamed milieus display increased expression of RAGE ligands; here demonstrated in allograft tissue.
In our studies, we observed that orthotopic transplantation of syngeneic grafts resulted in indefinite graft survival of at least 100 days. Thus, in the absence of pharmacological intervention, syngeneic grafts were highly compatible in the host. Furthermore, these considerations suggest that the impact, if any, of I/R damage and modulation of graft survival in this model was not substantive. Previous studies explored roles for RAGE in I/R injury in the isolated perfused heart (32). In that model, however, dramatic and sustained reduction of oxygenation in the heart was noted which resulted in extensive necrosis and reduction in organ function. Here, in this model, no evidence of significant I/R-induced damage was noted in syngeneic transplants. These considerations suggest that the chief factor driving recruitment of the ligand-RAGE axis in these cardiac allografts was the allorecognition stimulus.
It is important to note that these experiments in vivo were performed in the absence of agents established to confer immune suppression. Despite the absence of such agents in the treatment of mice undergoing allotransplantation, administration of sRAGE resulted in a significant increase in graft survival time. In parallel, the inflammatory infiltrates, particularly the numbers of T cells, were significantly decreased by RAGE blockade. Future studies must address the impact of combined therapy of sRAGE with standard immune suppressive agents. The importance of such studies lies in the fact that sRAGE has been administered to animals for up to 4 months without any apparent adverse consequences on survival, growth or homeostasis (33). In contrast, agents used to impart immune suppression often have multiple side effects; indeed, efforts to reduce the requirement for calcineurin inhibitors are one common strategy aimed to reduce toxicity in human transplantation (34,35). Based on our findings, we propose that it is reasonable to examine the impact of combined therapy in these alloimmune models.
In conclusion, these studies identify central roles for RAGE in inflammatory pathways triggered by complete allomismatch. Extension of our findings to human MLRs further supports testing RAGE antagonism in allogeneic transplantation, particularly in an effort to reduce toxicity and direct allograft injury imparted by certain immunosuppressive agents currently in use in clinical transplantation.
This work was funded by a grant from the United States Public Health Service (HL60901).