Although fully MHC-mismatched murine cardiac allografts are rapidly rejected, allografts mismatched at only MHC class I or class II alleles survive long term; the immunologic basis for the long-term survival of MHC class I- or II-mismatched allografts is unknown. We examined the roles of two recently described inhibitory receptors, B and T lymphocyte attenuator (BTLA) and programmed death-1 (PD-1), in the survival of partially or fully MHC-mismatched allografts using gene-deficient recipients as well as through use of blocking mAbs in wild-type hosts. Partially MHC-mismatched allografts showed strong induction of BTLA, but not PD-1 mRNA and survived long term in wild-type recipients, whereas targeting of BTLA or its ligand, herpesvirus entry mediator, but not PD-1, prompted their rapid rejection. By contrast, fully MHC-mismatched cardiac allografts were acutely rejected in wild-type recipients despite the induction of both BTLA and PD-1. Targeting of PD-1 in several fully MHC-mismatched models accelerated rejection, whereas targeting of BTLA unexpectedly enhanced PD-1 induction by alloreactive CD4 and CD8 T cells and prolonged allograft survival. In vitro studies using allogeneic dendritic cells and T cells showed that at low levels of T cell activation, BTLA expression was primarily induced, but that with increasing degrees of T cell activation, the expression of PD-1 was strongly up-regulated. These data suggest that BTLA and PD-1 exert distinct inhibitory actions in vivo, with the BTLA/herpesvirus entry mediator pathway appearing to dominate in regulating responses against a restricted degree of allogeneic mismatch.
Alloreactive T cells induce the rejection of MHC-mismatched transplants (1). Consistent with a requirement for appropriate costimulatory signals for optimal T cell activation, inhibition of CD28/B7 interactions can block acute rejection in various models of solid organ transplantation, although allografts are still usually lost as a result of residual immune responses that are manifested as chronic rejection. In recent years, additional CD28/B7 superfamily member pathways were identified, including ICOS/B7-related protein-1 (B7RP-1)3 and programmed death-1 (PD-1)/PD ligand-1 (PD-L1)/PD-L2, plus two B7 homologues, B7-H3 and B7-H4, whose receptors remain unknown (2, 3). Studies of these various CD28 and B7 homologue molecules in allograft models showed that their targeting alone or in conjunction with subtherapeutic doses of immunosuppression can promote long-term graft survival (4, 5, 6, 7, 8). Data from transplant models indicate that CD28/B7 superfamily members can be readily divided into those with stimulatory (CD28/B7, ICOS/B7RP-1, and B7-H3) or inhibitory (CTLA4/B7 and PD-1/PD-L1/PD-L2) properties (9).
B and T lymphocyte attenuator (BTLA) is a recently identified costimulatory receptor that is induced by activated CD4+ and CD8+ T cells and also expressed by B cells, macrophages, and bone marrow-derived myeloid dendritic cells (DC) (10, 11, 12). BTLA cross-linking causes tyrosine phosphorylation and association with the Src homology domain 2-containing protein tyrosine phosphatase-1 and -2 via dual ITIM motif (13). BTLA has key structural and functional similarities to the two other T cell inhibitory receptors, CTLA-4 and PD-1, and its ligation generates signals that dampen T and B cell activation in vitro and in vivo (10).
Although initial data suggested the B7 superfamily member, B7x (B7H4, B7S1) as a BTLA ligand (10, 14, 15), recent work shows the ligand for BTLA is actually the TNFR superfamily member, herpesvirus entry mediator (HVEM; TR2) (16). We have evaluated the role of the BTLA/HVEM pathway in experimental cardiac allograft rejection. Based on existing data, we reasoned that BTLA should act to reduce the strength of alloresponses and promote long-term engraftment. Indeed, our results confirm an inhibitory role for BTLA.
More importantly, we have identified in vivo conditions in which BTLA-HVEM interactions exert a regulatory role, particularly those responses directed against partially MHC-mismatched allografts. These results highlight an unexpected complexity in the regulation of alloimmunity and help to clarify the relative hierarchy between BTLA and PD-1 in regulating host CD4+ and CD8+ T cell responses in vivo in varying settings of alloactivation.
Materials and Methods
BTLA−/− (10), PD-1−/− (17), and dual BTLA−/− and PD-1−/− mice were backcrossed for more than eight generations on a C57BL/6 background; HVEM−/− mice were generated by homologous recombination and backcrossed more than five generations on a B6 background (Q. Ye et al., manuscript in preparation). Wild-type C57BL/6 (H-2b), BALB/c (H-2d), C57BL6/DBA F1 (H-2b/d), Bm12 (B6.C-H2bm12/KhEg), and Bm1 (B6.C-H2bm1/ByJ) mice were purchased from The Jackson Laboratory, housed in specific pathogen-free conditions, and used for studies approved by the institutional animal care and use committee of Children’s Hospital of Philadelphia.
Quantitative PCR (qPCR)
We performed qPCR as previously described (19). Briefly, RNA was extracted with TRIzol (Invitrogen Life Technologies), RT of random hexamers was performed with an ABI PRISM 5700 unit (Applied Biosystems), and specific primer and probe sequences for target genes were used for qPCR amplification of total cDNA (TaqMan PDAR; Applied Biosystems). Relative quantitation of target cDNA was determined using a control value of 1; the sample cDNA content was expressed as the fold change from the control value. Differences in cDNA input were corrected by normalizing signals obtained with specific primers to ribosomal RNA; nonspecific amplification was excluded by performing RT-PCRs without target cDNA.
Alloreactive T cells were generated by i.v. injection of 40 × 106 CFSE-labeled B6 spleen and lymph node cells into B6/DBA F1 recipients, a parent→F1 MHC mismatch in which only donor cells respond (18). Splenocytes harvested after 3 days were incubated with CD4-PE, CD8-PE, CD25-PE, CD44-PE, CD62L-PE, PD-1-PE, ICOS-PE, and biotin-conjugated anti-H-2Kd and anti-H-2Dd mAb. Donor alloreactive T cells were identified by gating on H-2Kd and H-2Dd cells (FACSCalibur; BD Biosciences), and their proliferation was assessed by CFSE division profiles (18). For intracellular cytokine staining, splenocytes (3 × 106/ml) were treated with Golgi-Stop (BD Pharmingen), stimulated for 4 h with PMA (3 ng/ml) and ionomycin (1 μM) in 24-well plates in complete medium (RPMI 1640, 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 μM 2-ME), and stained with cell surface markers (CD4-PE or CD8-PE, biotin-conjugated H-2Kd or H-2Dd, followed by streptavidin-PerCP), fixed, and stained with IFN-γ-allophycocyanin or IL-2-allophycocyanin after permeabilization (Perm-Wash buffer; BD Pharmingen).
In vitro cellular assays
For propagation of bone marrow-derived DC, bone marrow cells harvested from the femurs and tibia were cultured for 5–7 days in 24-well plates (2 × 106/well) in medium plus mouse GM-CSF (5 ng/ml) and IL-4 (10 ng/ml) (20, 21). One-way MLR cultures were performed in triplicate, using magnetic column-eluted splenic T cells (2 × 105/well) as responders and gamma-irradiated (20 Gy) DC as stimulators. Cultures were maintained in complete medium for 72–96 h, and T cell proliferation was determined by BrdU incorporation or CFSE dilution profile. BrdU staining with a BrdU labeling kit (BD Pharmingen) was performed using the manufacturer’s instructions. Cells were pulsed with BrdU, treated with FcR-blocking CD16/CD32 mAbs, stained with cell surface markers, fixed, permeabilized, treated with DNase/Triton X-100, stained with anti-BrdU mAb, and analyzed by flow cytometry.
Immunospot assays for IFN-γ were performed by coating ELISPOT plates (BD Pharmingen) with anti-IFN-γ mAb, blocking, and addition of responder cells isolated from cardiac transplant recipients plus donor splenocytes or bone marrow-derived DC as stimulators; recipient splenocytes or DC were used as syngeneic controls. At 24 h, cells were discarded, and wells were washed, followed by biotinylated anti-IFN-γ mAb, streptavidin-HRP, and substrate. Spots were counted using an Immunospot Analyzer (Cellular Technology), and recipient anti-donor responder frequency was determined as the number of IFN-γ spot-forming cells per 106 splenocytes (22).
Grafts were sonicated in lysis buffer containing Triton X-100 and protease inhibitors, followed by centrifugation and assay of supernatant protein content. Proteins were reduced, separated by SDS-PAGE, and transferred to nitrocellulose membranes. Membranes were blocked, incubated with primary and HRP-linked secondary Abs, and, after the substrate reaction, analyzed using National Institutes of Health Image.
Intra-abdominal vascularized cardiac allografting was performed as previously described (4) using 6- to 8-wk-old mice. Briefly, donor ascending aorta and pulmonary artery were anastomosed end-to-side to recipient infrarenal aorta and inferior vena cava, respectively. Graft survival was assessed twice daily by abdominal palpation; rejection was defined as total cessation of cardiac contraction and was confirmed by histology.
Portions of harvested allografts were fixed in formalin, paraffin-embedded or snap-frozen, and analyzed by immunoperoxidase staining with mAbs and an Envision kit (DakoCytomation) (4).
Allograft survival was used to generate Kaplan-Meier survival curves, and comparison between groups was performed by log-rank analysis.
BTLA and HVEM regulate acceptance of partially MHC-mismatched cardiac allografts
Primarily vascularized cardiac allografts are the most frequent organ transplant undertaken in mice and may be performed across full MHC disparities, with rejection in 7–8 days, or across MHC class I or II disparities, which leads to long-term survival (>100 days) (23). The basis for this unexpectedly long-term survival of cardiac transplants across partial MHC disparities is unknown and has received little attention. As anticipated from the literature, we indeed found that cardiac allografts performed across an MHC class II mismatch (Bm12→B6) survived long term in wild-type recipients (mean survival time (MST), >100 days; n = 6). Histologic assessment of these allografts harvested at 2 wk after transplant showed preservation of myocardial architecture and generally only sparse mononuclear cell infiltration (Fig. 1,a). In contrast, BTLA−/− recipients rejected Bm12 cardiac allografts by 2–3 wk after transplant (MST, 14.3 ± 3.8 days; n = 12; p < 0.001), and histology showed a marked increase in leukocyte infiltration and myocardial injury (Fig. 1,a). In addition, comparable abrogation of Bm12 allograft survival was seen with mAb targeting of BTLA in wild-type recipients (MST, 23.2 ± 3.2; n = 6; p < 0.001) or by engraftment of recipients lacking the BTLA ligand, HVEM (MST, 17.4 ± 4.2 days; n = 8; p < 0.001; Fig. 1,a). Thus, BTLA and HVEM are required to allow long-term survival of partially mismatched cardiac allografts. In contrast to results obtained with BTLA−/− recipients, PD-1−/− recipients receiving Bm12 cardiac allografts exhibited an 80% long-term allograft survival (Fig. 1,b), although we did observe a minor role for PD-1 in regulating responses to Bm12 cardiac allografts. Dual BTLA−/− and PD-1−/− knockout mice (DKO) mice rejected Bm12 donor hearts more rapidly (MST, 10.5 ± 1.5 days; n = 4) than singly deficient BTLA−/− recipients (p < 0.05) or wild-type controls (p < 0.0001; Fig. 1 b).
Like MHC class II-mismatched grafts, MHC class I-mismatched (Bm1→B6) cardiac allografts survived long term when transplanted to wild-type B6 mice, but were rejected in BTLA−/− mice (Fig. 1,c). Furthermore, in contrast to wild-type B6 recipients, the MHC class I-mismatched allografts in BTLA−/− recipients showed increased mononuclear cell infiltration and progressive tissue damage indicative of the development of cellular rejection (Fig. 1 c). PD-1−/− recipients receiving Bm1 cardiac allografts had 100% long-term allograft survival (data not shown). Collectively, these findings indicate that BTLA, in contrast to PD-1, is capable of inhibiting the generation of a functional allogeneic immune response in the context of partial MHC mismatches.
BTLA suppresses MHC class II-dependent T cell responses
The unexpected rejection of Bm12 allografts by BTLA−/−, but not PD-1−/−, mice suggested that BTLA and PD-1, or their ligands, might be differentially expressed in partially MHC-mismatched allografts. BTLA mRNA expression within Bm12 allografts was ∼20-fold higher than PD-1 at 7 days after transplant, whereas no BTLA expression was detected within Bm12 hearts engrafted into BTLA−/− recipients, indicating BTLA expression primarily by infiltrating host leukocytes (Fig. 2,a). Comparable BTLA expression was observed within long-surviving allografts (data not shown). Unlike BTLA, only very low levels of PD-1 were detected in Bm12 allografts in either wild-type or BTLA−/− recipients (Fig. 2,a). No differences in the levels of expression of HVEM, PD-L1, or PD-L2 were seen between wild-type and BTLA−/− recipients (Fig. 2 a). These data suggest that in the Bm12→B6 model, BTLA is the predominant inhibitory receptor expressed by infiltrating alloreactive T cells, and that in the absence BTLA, there is no compensatory increase in expression of additional inhibitory molecules.
We next studied the in vitro and in vivo responses of T cells from wild-type and BTLA−/− mice to MHC class II Ags. First, we examined the in vitro proliferation of purified wild-type or BTLA−/− CD4+ T cells cocultured with irradiated Bm12 DC. Proliferation of BTLA−/− T cells was increased compared with that of wild-type T cells, as measured by either BrdU incorporation (Fig. 2,b) or CFSE dilution (Fig. 2,c). To assess in vivo responses, 40 million CFSE-labeled wild-type or BTLA−/− splenocytes were adoptively transferred into irradiated Bm12 hosts, and donor CD4+ T cell proliferation was assessed. Although a large portion of wild-type CD4+ T cells remained undivided 72 h after adoptive transfer, almost all BTLA−/− CD4+ T cells had entered the cycle and proceeded through several rounds of division (Fig. 2 d). Hence, BTLA regulates CD4+ T cell alloactivation and proliferative responses to MHC class II Ags.
MHC class II-restricted CD4+ T cell proliferation dominates host alloresponses in the Bm12→B6 model, although host responses are known to include stimulation of CD8+ precursor CTL by class II-restricted CD4+ T cells (24, 25, 26). We found that although proliferative responses of CD8+ T cells in irradiated Bm12 hosts were low compared with those of CD4 cells, the alloactivation and proliferation of CD8+ T cells from BTLA−/− mice were marginally increased over control cells in this assay system (Fig. 2,e). We examined recipient anti-donor responder frequencies by ELISPOT, with the readout of IFN-γ spot-forming cells by recipient splenocytes (22). BTLA−/− recipient splenocytes had significantly higher anti-donor responder frequencies when challenged with Bm12 APCs (Fig. 2 f), consistent with the increased allogeneic proliferation in vitro and the accelerated graft rejection in vivo of T cells from BTLA−/− mice.
Minor role of BTLA in fully MHC-mismatched alloresponses
We next tested whether BTLA played a similar dominant role in regulating responses to fully MHC-mismatched cardiac allografts as it did for partially MHC-mismatched cardiac allografts. Wild-type recipients (B6, H-2b) rejected cardiac grafts (BALB/c, H-2d) in 7–10 days (MST, 8 ± 1 days; n = 6), whereas BTLA−/− recipients showed a small and unexpected prolongation of graft survival (MST, 12 ± 5 days; n = 6; p < 0.05; Fig. 3,a). In addition, wild-type mice treated with a neutralizing anti-BTLA mAb showed a similar prolongation of allograft survival (MST, 13 ± 1 days; n = 4; p < 0.05) compared with control IgG treated recipients (MST, 8 ± 1 days; n = 4; Fig. 3,b). Furthermore, addition of a subtherapeutic course of rapamycin prolonged graft survival in wild-type mice by a few days (MST, 11 ± 2 days; n = 6; p < 0.05), but significantly prolonged graft survival in BTLA−/− mice (MST, 53 ± 12 days; n = 8; p < 0.001), with 25% of the latter recipients achieving long-term acceptance (Fig. 3 c). Hence, in the case of fully MHC-mismatched cardiac allografts, loss of BTLA did not accelerate allograft rejection, but, rather, caused a surprising, albeit small, increase in allograft survival. By contrast, the presence or the absence of BTLA had no effect on the tempo of rejection of B6 cardiac allografts by BALB/c recipients; all allografts were rejected within 7–10 days (n = 4/group; p > 0.05).
To understand the prolongation of fully MHC-mismatched graft survival, we measured the expression of cytokines and chemokine receptors important to host T cell recruitment in this model (27, 28), using allografts harvested 7 days after transplant. We found decreases in IL-2 and IFN-γ mRNA in BTLA−/− recipients compared with wild-type recipients (Fig. 3,d). We also found reduced expression of CXCR3 and CCR5 in BTLA−/− recipients compared with wild-type recipients (Fig. 3,d). Therapy with rapamycin accentuated differences in cytokine and chemokine receptor mRNA expression between BTLA−/− and wild-type recipients (Fig. 3,d). Given a key role for IFN-γ-induced IFN-inducible protein 10 (IP-10) production in promoting CXCR3+ cell recruitment and allograft rejection in this model (29), we performed Western blotting, which confirmed that allografts in BTLA−/− recipients had reduced IP-10 and CXCR3 proteins compared with wild-type controls, with or without rapamycin therapy (Fig. 3 e).
To assess whether the lack of BTLA affected the strong alloactivation and proliferation induced in T cells by 72 h in this model (19), we used the parent-to-F1 model involving transfer of CFSE-labeled cells across fully allogeneic barriers (Fig. 3,f). In this model, the activation and cell cycle progression of CD4+ responses were similar for BTLA−/− and wild-type cells, and CD8+ T cells from BTLA−/− mice were only marginally decreased compared with those from wild-type controls (Fig. 3,f). However, the evaluation of intracellular cytokine production by alloreactive T cells showed decreased IL-2 and IFN-γ production by alloreactive BTLA−/− CD4+ and CD8+ T cells compared with wild-type T cells (Fig. 3,g). Again, a subtherapeutic rapamycin dose caused a modest decrease in proliferation of BTLA−/− T cells compared with wild-type T cells, particularly CD8+ responses (Fig. 3,f), and decreased production of IL-2 and IFN-γ by both T cell subsets (Fig. 3 g). These data indicate that T cell activation, proliferation, and production of cytokines such as IL-2 and IFN-γ are decreased in BTLA−/− mice, especially when recipients are treated with limited immunosuppression, and that these impaired responses are associated with modulation of chemokine/chemokine receptor effector pathways.
Involvement of PD-1 and BTLA in fully MHC-mismatched alloresponses
In considering explanations for the differing effects of BTLA in the partial MHC-mismatch and full MHC mismatch models, we wondered whether differential reliance on PD-1 between these models might play a role. Therefore, we examined the contributions of both PD-1 and BTLA in the fully MHC-mismatched model (Fig. 4). We found, first, that BALB/c cardiac allografts were rejected at similar rates (p > 0.05) by C57BL/6 wild-type mice and DKO mice (Fig. 4,a). Second, consistent with the DKO data, mAb blockade of PD-1 increased the rate of rejection of fully MHC-mismatched allografts by BTLA−/− recipients (p < 0.001; Fig. 4,b). Third, the duration of allograft survival in BTLA−/− recipients receiving subtherapeutic course of rapamycin (MST, 53 ± 12 days; n = 8) was markedly decreased by loss of PD-1, as seen by examining either DKO recipients (MST, 12.8 ± 2.2 days; n = 4; p < 0.001; Fig. 4,c) or by mAb Ab blockade of PD-1 in BTLA−/− mice (MST, 14.0 ± 3.5 days; n = 4; p < 0.001; Fig. 4 d). In summary, in contrast to partial MHC-mismatched allografts, the responses against fully MHC-mismatched cardiac allografts are regulated by both BTLA and PD-1.
We next asked whether PD-1 regulated the proliferation and function of T cells responding to fully MHC-mismatched allografts. Analysis by qPCR of BALB/c cardiac allografts harvested on day 7 after transplant from C57BL/6 recipients showed intragraft expression of BTLA, PD-1, and their ligands, HVEM, PD-L1, and PD-L2 (Fig. 5,a). By comparison with wild-type recipients, BALB/c allografts harvested from BTLA−/− recipients had increased PD-1 expression (Fig. 5 a; p < 0.01). In contrast to PD-1 expression, the expression of HVEM, PD-L1, and PD-L2 was not increased in BTLA−/− recipients. These results suggest that in the absence of BTLA, host leukocytes might express more PD-1 in response to allostimulation.
To directly examine PD-1 expression by alloreactive wild-type or BTLA−/− T cells, we adoptively transferred CFSE-labeled splenocytes into irradiated Bm12 (class II-mismatched) or B6D2F1 (fully MHC-mismatched) recipients. At analysis 72 h later, we found that in the MHC class II partial mismatch, PD-1 was weakly expressed by alloreactive CD4+ T cells, but not at all by CD8+ T cells, from wild-type or BTLA−/− mice (Fig. 5,b). In contrast, with a full MHC mismatch, PD-1 expression by both CD4+ and CD8+ donor T cells was markedly increased, and the extent of PD-1 expression was higher in BTLA−/− vs wild-type T cells (Fig. 5,b). Moreover, treatment with rapamycin reduced PD-1 expression by wild-type T cells, but had only minor effects on PD-1 induction by T cells from BTLA−/− mice (Fig. 5 b).
Lastly, we used in vivo and in vitro approaches to examine the roles of BTLA and PD-1 in regulating T cell proliferation and cytokine production in response to fully MHC-mismatched allostimulation (Fig. 5, c–e). Compared with wild-type or BTLA−/− cells, DKO cells showed enhanced proliferation (Fig. 5,c) and Th1 cytokine production (Fig. 5,d). Therapy with rapamycin decreased the alloactivation-induced proliferation (Fig. 5,c) and cytokine production (Fig. 5,e) of CD4+ and CD8+ T cells from wild-type and BTLA−/− donors, but did not block these events in DKO CD4+ or CD8+ T cells (Fig. 5, c and e). Indeed, the production of IL-2 and IFN-γ was increased in DKO T cells compared with wild-type and BTLA−/− T cells (Fig. 5,d), including in the presence of rapamycin therapy (Fig. 5,e). Collectively, these data indicate that 1) PD-1 expression is highly induced on the surfaces of alloreactive CD4+ and CD8+ T cells upon exposure to fully MHC-disparate allografts; 2) the levels of PD-1 on alloreactive CD4+ and CD8+ T cells are still further increased in the absence of BTLA; and 3) increased PD-1 expression is associated with inhibitory effects on the alloantigen-induced production of cytokines such as IL-2 and IFN-γ. In associated in vitro studies, as T cell activation increased in response to allogeneic DC (Fig. 6), the induction of PD-1 was increasingly apparent compared with that of BTLA. BTLA up-regulation occurred upon T cell activation, but did not show expansion comparable with that of PD-1 with increasing T cell activation, suggesting that the strength of T cell activation determines the relative importance of these two pathways.
This study identifies a unique action of the BTLA/HVEM pathway in regulating in vivo allogeneic responses that may have therapeutic potential. BTLA is the most recently identified inhibitory immunoreceptor expressed by lymphocytes (10). Two others, CTLA-4 and PD-1 (30, 31), have been shown to exert important inhibitory actions relevant for maintaining tolerance (32), but analysis of the in vivo role of BTLA has been limited. Neonatal CTLA4−/− mice show uncontrolled lymphocyte proliferation, infiltration of multiple organs, and death by 3–4 wk of age (33, 34). Adult PD-1−/− mice develop glomerulonephritis (17), lupus-like arthritis (17), or autoimmune cardiomyopathy (35). In contrast, in vivo findings for BTLA−/− to date have been restricted to an increased susceptibility to experimental allergic encephalomyelitis, consistent with an inhibitory action of BTLA, but the effect was quantitative, not qualitative (10). The current results demonstrate a qualitative requirement for BTLA in the tolerance of partially MHC-mismatched cardiac allografts in vivo, an action unique to BTLA and distinct from that of PD-1.
MHC class II-mismatched cardiac allografts survive permanently (>100 days), but acquire a progressive vasculopathy and other hallmarks of chronic rejection; indeed, this is currently the standard model of chronic rejection in mice (36, 37). Consistent with a recent report from Koga et al. (38), we found that targeting the PD-1 pathway could not restore acute rejection in this model. However, surprisingly, the survival of such MHC class II-mismatched allografts is completely dependent on BTLA and its ligand, HVEM. We found that BTLA−/− mice, wild-type mice treated with a neutralizing anti-BTLA mAb, or HVEM−/− mice exhibit dramatically accelerated rejection of MHC class II-mismatched allografts, with extensive lymphocyte infiltration, progressive myocardial injury, and 100% graft loss within 2–3 wk of transplantation. Correspondingly, BTLA, but not PD-1, was significantly induced on alloreactive T cells in this setting of partial MHC mismatch, and in the absence of BTLA or HVEM, T cell responses to MHC class II-mismatched allogeneic cells were markedly enhanced. Lack of BTLA also accelerated the rejection of MHC class I-mismatched cardiac allografts. Thus, BTLA and HVEM are the major regulators of in-host allogeneic responses to class I- or class II-mismatched allografts. Overexpression of HVEM on APCs was previously shown to inhibit peptide Ag-specific T cell proliferation in wild-type, but not BTLA−/−, mice, but only when the peptide concentration was low; no effect was observed with increased Ag load (16). This is consistent with our assessment that BTLA exerts its main effect when the strength of the immune response is weak.
Unlike alloresponses to partial MHC mismatches in the mouse, transplantation across a full MHC mismatch elicits one of the most potent immune responses known (23, 36, 37). In our studies we observed that responses to fully MHC-mismatched grafts were associated with increased expression of BTLA, PD-1, and their ligands. Surprisingly, alloreactive T cells from BTLA−/− mice displayed decreased proliferation and decreased cytokine production in response to fully allogeneic cells in vivo, and this response was further attenuated by a subtherapeutic regimen of rapamycin. Moreover, BTLA−/− mice had slightly prolonged cardiac allograft survivals compared with wild-type controls, and again, rapamycin therapy enhanced the effects of BTLA targeting. At face value, these data could either indicate an unexpected positive costimulatory function for BTLA in fully MHC-mismatched alloresponses, in contrast to previous data (10), or reflect the increased PD-1 expression found under these conditions in BTLA−/− mice.
We favor the latter interpretation. PD-1 was markedly induced by alloreactive CD4+ and CD8+ T cells upon exposure to fully allogeneic cells and served to dampen T cell proliferation and Th1 cytokine production. We have previously shown that PD-1 and its ligands, PD-L1 and PD-L2, are all expressed within cardiac transplants in this fully allogeneic model (5). The inhibitory effects of the PD-1/PD-L1/PD-L2 pathway are usually insufficient to control development of effector T cell responses and allograft rejection given the induction and functions of multiple costimulatory molecules after TCR ligation in this system, including CD28/B7, CD40L/CD40, ICOS/B7RP-1, and others (9). However, increasing inhibitory signals by stimulation of the PD-1 receptor using PD-L1.Ig in conjunction with a subtherapeutic course of immunosuppression, either cyclosporine or rapamycin, can markedly prolong or induce permanent allograft survival in this model (5).
In the current context, increased signaling by enhancing the levels of PD-1 expression might also be expected to facilitate graft survival. This is precisely what was found in BTLA−/− mice, in which BTLA−/− recipients of fully allogeneic hearts had a modestly increased level of PD-1 induction and modestly enhanced allograft survival. Likewise, BTLA−/− mice engrafted with fully allogeneic hearts and treated with rapamycin showed higher levels of PD-1 than wild-type controls and correspondingly still greater prolongation of allograft survival. The key role of PD-1 in these events was shown by the quick restoration of allograft rejection in DKO mice or in BTLA−/− recipients treated with an anti-PD-1 mAb. These data indicate that BTLA can regulate various T cell responses, including induction of PD-1, with functional consequences in allograft models. Moreover, our in vitro studies supported the concept that the relative expression of BTLA and PD-1 varies depending upon the extent of T cell activation upon exposure to allogeneic cells.
In summary, our data show that BTLA and HVEM constitute a key negative costimulatory pathway that regulates host T cell responses to allogeneic class I or class II Ags; by comparison, PD-1 plays only a minor role in such responses. However, in the context of fully MHC-mismatched Ags, although both BTLA and PD-1 are induced, PD-1 expression plays the dominant role, and indeed, BTLA functions as an inhibitor of PD-1 induction. These findings are of possible clinical significance, because chronic rejection of partially MHC-mismatched organ transplants represents a currently unresolved Achilles’ heel of human transplantation. One implication of our studies is that potentiation of the inhibitory actions of BTLA might help promote long-term graft survival and prevent chronic rejection, especially in the context of partially MHC-mismatched donor/recipient combinations.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by National Institutes of Health Grant AI40152 (to W.W.H.).
Abbreviations used in this paper: B7RP-1, B7-related protein-1; PD-1, programmed death-1; BTLA, B and T lymphocyte attenuator; DC, dendritic cell; DKO, doubly deficient in PD-1 and BTLA; HVEM, herpesvirus entry mediator; IP-10, IFN-inducible protein 10; MST, mean survival time; PD-L1, PD ligand-1; qPCR, quantitative PCR.