Alloantigen expression on non-hematopoietic cells reduces graft-versus-leukemia effects in mice

Allogeneic hematopoietic stem cell transplantation (HSCT) is used effectively to treat a number of hematological malignancies. Its beneficial effects rely on donor-derived T cell–targeted leukemic cells, the so-called graft-versus-leukemia (GVL) effect. Induction of GVL is usually associated with concomitant development of graft-versus-host disease (GVHD), a major complication of allogeneic HSCT. The T cells that mediate GVL and GVHD are activated by alloantigen presented on host antigen-presenting cells of hematopoietic origin, and it is not well understood how alloantigen expression on non-hematopoietic cells affects GVL activity. Here we show, in mouse models of MHC-matched, minor histocompatibility antigen–mismatched bone marrow transplantation, that alloantigen expression on host epithelium drives donor T cells into apoptosis and dysfunction during GVHD, resulting in a loss of GVL activity. During GVHD, programmed death–1 (PD-1) and PD ligand–1 (PD-L1), molecules implicated in inducing T cell exhaustion, were upregulated on activated T cells and the target tissue, respectively, suggesting that the T cell defects driven by host epithelial alloantigen expression might be mediated by the PD-1/PD-L1 pathway. Consistent with this, blockade of PD-1/PD-L1 interactions partially restored T cell effector functions and improved GVL. These results elucidate a previously unrecognized significance of alloantigen expression on non-hematopoietic cells in GVL and suggest that separation of GVL from GVHD for more effective HSCT may be possible in human patients.

Donor immunity in allogeneic hematopoietic stem cell transplantation (HSCT) harnesses beneficial graft-versus-leukemia (GVL) effects; therefore, allogeneic HSCT represents a very potent form of immunotherapy for hematological malignancies (1, 2). Induction of GVL is usually associated with the development of graft-versus-host disease (GVHD), which is a major complication after allogeneic HSCT. T cell depletion of the donor inocula prevents GVHD and leads to a loss of the GVL effect (3–5). Both GVL and GVHD are mediated by donor T cells, which recognize alloantigens presented on host APCs (6, 7). Donor CTLs and inflammatory cytokines are major effectors of GVHD, whereas CTLs are primarily responsible for GVL (8, 9). In patients with advanced-stage leukemia and lymphoma, relapse is still a major cause of mortality after allogeneic HSCT even after the development of severe GVHD. Thus, improvements in our understanding of the pathophysiology of GVHD and GVL are urgently needed to develop more effective therapies for malignant diseases.

Alloantigens are expressed on the three major components in HSCT recipients in the context of GVHD and GVL: hematopoietically derived APCs, GVHD target epithelium, and leukemia cells. Several studies have shown that host APCs are crucial for the induction of both GVHD and GVL (6, 7, 9–11). Alloantigen expression on epithelium is also critical for the induction of GVHD in MHC-matched, minor histocompatibility antigen–mismatched (mHA-mismatched) models of bone marrow transplantation (BMT) (10), but GVHD can occur in the absence of alloantigen expression on epithelium in MHC-mismatched models of BMT (9). However, the effect of alloantigen expression on non-hematopoietic cells such as the epithelium in GVL is not well defined. In this study, we addressed this important issue in mHA-mismatched models of BMT.

Alloantigen expression on host non-hematopoietic cells augments acute GVHD but reduces GVL effects. We generated BM chimeric mice that express alloantigens on APCs, which are essential for the induction of both GVHD and GVL (6, 7, 12). BM chimeras were created by reconstituting lethally irradiated C3H.Sw (C3: H-2^b) mice with 5 × 10⁶ T cell–depleted (TCD) BM cells isolated from C57BL/6 (B6, H-2^b) mice that differ from C3 mice at multiple mHAs ([B6→C3] chimeras). Control chimeras, [B6→B6], were identically created. Four months later, donor repopulation of hematopoiesis was confirmed by flow cytometry as shown previously (6, 9, 12). Thus, [B6→C3] chimeric mice expressed B6-derived mHAs on hematopoietically derived APCs but not on non-hematopoietic target cells. In contrast, [B6→B6] mice expressed B6-derived mHAs on both APCs and target epithelium. These chimeras were used as BMT recipients; they were reirradiated and injected with 5 × 10⁶ TCD BM cells alone or with various doses of CD8⁺ T cells from C3 donors. After BMT, GVHD mortality was higher in [B6→B6] mice than in [B6→C3] mice (Figure 1A). Clinical GVHD scores (13) in surviving animals were also higher in [B6→B6] mice than in [B6→C3] mice (Figure 1B). Mortality and morbidity from GVHD in [B6→C3] mice were almost equivalent to those in [B6→B6] mice given a 1-log lower T cell dose. This finding confirmed the previous observation of a lack of alloantigen expression on host epithelium significantly reducing GVHD across mHA disparity (10). We then tested the effect of alloantigen expression on GVHD target epithelium on GVL effects. These chimeric mice were transplanted as described above together with 2,500 B6-derived EL4 cells as a model of residual leukemia after BMT. As expected, 100% of both types of chimeric mice that received TCD BM cells died from leukemia by day +20 after BMT (Figure 1C), whereas leukemia-free survival was significantly prolonged in mice that received donor T cells, demonstrating a significant GVL effect. However, this GVL effect was not potent in [B6→B6] mice, and all mice subsequently died from leukemia. Surprisingly, leukemia mortality was significantly lower in [B6→C3] mice that did not express alloantigens on their non-hematopoietic cells (62% vs. 100%; P < 0.05). GVL effects in [B6→B6] mice appeared to be almost equivalent to those in [B6→C3] mice given a 1-log lower T cell dose.

Figure 1

Alloantigen expression on host non-hematopoietic cells augments acute GVHD but reduces GVL effects. (A–C) [B6→C3] (diamonds) and [B6→B6] chimeras (triangles) were created by reconstituting lethally irradiated C3 and B6 mice with 5 × 10⁶ TCD BM cells from B6 mice. Four months later, the chimeras were reirradiated and injected with 5 × 10⁶ TCD BM cells alone (open symbols) or with 1 × 10⁶ (black symbols) or 0.1 × 10⁶ (gray symbols) CD8⁺ T cells from C3 donors (as indicated in parentheses × 10⁶). Survival (A) and clinical GVHD scores (B) after BMT (n = 3–8/group). (C) Leukemia mortality after BMT in chimeras injected with EL4 cells (n = 5–21/group). Data from 3 similar experiments were combined. (D–F) [Db→Ba] (diamonds) and [Db→Db] (triangles) chimeras were reirradiated and injected with TCD BM alone (open symbols) or with 2 × 10⁶ T cells from Ba donors (filled symbols). Survival (D) and clinical GVHD scores (E) after BMT from a representative experiment of 2 similar experiments (n = 4–7/group). (F) Leukemia mortality after BMT in mice injected with P815 cells. Data from 2 similar experiments were combined (n = 6–18/group). (G–I) [Ba→Db] (diamonds) and [Ba→Ba] (triangles) chimeras were similarly transplanted with 5 × 10⁶ TCD BM cells alone (open symbols) or with 2 × 10⁶ T cells from Db donors (filled symbols). Survival (G) and clinical scores (H) after BMT (n = 3–10/group). (I) Leukemia mortality after BMT in chimeras injected with A20 cells (n = 5–10/group). Data from 2 similar experiments were combined. Clinical scores are shown as the mean ± SEM. *P < 0.05 compared with allogeneic controls.

We further confirmed these observations in a different strain combination: BALB/c (Ba, H-2^d) and DBA/2 (Db, H-2^d) mice that differed at multiple mHAs from each other. [Db→Ba] and control [Db→Db] chimeras were lethally reirradiated and injected with 5 × 10⁶ TCD BM cells alone or with 2 × 10⁶ Ba T cells. Mortality (Figure 1D, P = 0.08) and morbidity from GVHD (Figure 1E, P < 0.05) were higher in [Db→Db] mice than in [Db→Ba] mice. When cells were transplanted together with 2,000 Db-derived P815 cells, leukemia mortality was significantly lower in [Db→Ba] mice than in [Db→Db] mice (10% vs. 60%; P < 0.05) (Figure 1F).

Similar results were obtained when [Ba→Db] and control [Ba→Ba] chimeras were transplanted with 5 × 10⁶ TCD BM cells with or without 2 × 10⁶ Db T cells. In [Ba→Db] recipients, in which non-hematopoietic cells do not express alloantigens, mortality (Figure 1G, P = 0.08) and morbidity of GVHD (Figure 1H, P < 0.05) were lower, but GVL effects against Ba-derived A20 lymphoma cells were significantly more potent as compared with [Ba→Ba] controls (leukemia mortality: 30% vs. 100%; P < 0.05) (Figure 1I). Taken together, these results demonstrate that GVHD is decreased but GVL activity is enhanced in the absence of alloantigen expression on non-hematopoietic cells.

Alloantigen expression on non-hematopoietic cells enhances apoptosis and dysfunction of alloreactive T cells. GVHD and GVL in the C3 and B6 strain combination is dependent on donor CD8⁺ T cells (12, 14). To elucidate the mechanisms responsible for the enhancement of the GVL effect in [B6→C3] chimeric mice, which lack alloantigen expression on non-hematopoietic cells, the kinetics of donor CD8⁺ T cell expansion and activation were evaluated after BMT. Expansion of donor CD8⁺ T cells identified as CD5.1⁺CD8⁺ cells peaked on day +14 in the spleens of allogeneic [B6→B6] recipients and decreased thereafter (Figure 2A), as previously shown in this model (15). CD8 expansion was significantly greater in [B6→C3] mice than in [B6→B6] mice on days +14 and +21. We next assessed donor T cell apoptosis as a determinant of the kinetics of T cell expansion. Frequencies of annexin V⁺ apoptotic donor CD8⁺ T cells were significantly greater in the spleen of [B6→B6] mice as compared with that of [B6→C3] mice on day +14 (Figure 2B). Notably, surviving donor CD8⁺ T cells were significantly less activated in [B6→B6] mice than in [B6→C3] mice when evaluated based on the expression of CD69 (Figure 2C) and intracellular IFN-γ (Figure 2D) on annexin V^– donor CD8⁺ T cells. We next evaluated CTL activity in donor T cells isolated from the spleen on day +14 after BMT. CTL activity against EL4 targets was significantly reduced in the splenocytes of [B6→B6] mice as compared with [B6→C3] mice (Figure 2E). These results suggest that alloantigen expression on non-hematopoietic cells induces apoptosis and dysfunction of alloreactive T cells.

Figure 2

Alloantigen expression on host non-hematopoietic cells enhances the apoptosis and dysfunction of alloreactive T cells. [B6→C3] (diamonds and black bars) and [B6→B6] (triangles and gray bars) chimeras were transplanted as indicated in the legend for Figure 1. Syngeneic controls were [B6→B6] recipients of B6.Ly5.1 (CD45.1⁺) cells (open circles and white bars). (A) Numbers of donor CD8⁺ T cells in spleens. (B) Frequencies of annexin V⁺ donor CD8⁺ T cells. (C) Numbers of annexin V^– donor CD69⁺CD8⁺ T cells. (D) Numbers of annexin V^– IFN-γ–producing donor CD8⁺ T cells. (E) CTL activity against EL4. (B–E) Analysis was performed 14 days after BMT (n = 3–8/group). Representative data from 1 of the experiments are shown as the mean ± SD. *P < 0.05 compared with allogeneic controls.

Absence of alloantigen expression on host non-hematopoietic cells restores GVL effects. Self-recognition in the periphery facilitates the reactivity of mature T cells to foreign antigens (16). Therefore, it is possible that the expression of syngeneic MHC molecules and not the absence of alloantigens on non-hematopoietic cells may be responsible for the enhancement of the GVL effect in [B6→C3] chimeras. This possibility was tested in B6-background β₂m^–/– mice. [B6→β₂m^–/–] chimeras lacking functional MHC class I molecules on non-hematopoietic cells did not develop GVHD after transplantation with CD8⁺ T cells from C3 donors, as shown previously (17) (Figure 3A). In these mice, however, leukemia mortality was significantly delayed even in the absence of GVHD as compared with [B6→B6] recipients (Figure 3B, P < 0.05). The expansion and CTL activity of donor CD8⁺ T cells was significantly greater in [B6→β₂m^–/–] recipients than in [B6→B6] recipients (Figure 3, C and D). These results confirm that alloantigen expression on host epithelium induces apoptosis and dysfunction of alloreactive T cells, which results in impaired GVL effects.

Figure 3

Absence of alloantigen expression on host non-hematopoietic cells restores GVL effects. [B6→B6] (triangles) and [B6→β2m^–/–] (squares) mice were reirradiated and injected with 5 × 10⁶ TCD BM cells alone (open symbols) or with 1 × 10⁶ CD8⁺ T cells from C3 donors (filled symbols). (A) Survival after BMT. (B) Leukemia mortality in chimeras injected with EL4 cells (n = 6–9/group). Data from a representative experiment of 2 similar experiments are shown. Mean ± SEM numbers of donor CD8⁺ T cells in spleens (n = 3–6/group) (C) and CTL activity against EL4 (D). *P < 0.05 compared with allogeneic controls.

Alloantigen expression on host non-hematopoietic cells stimulates programmed death-1 and its ligand pathway. Programmed death–1 (PD-1) is a negative regulator of activated T cells and regulates T cell exhaustion during chronic infections (18–20). PD-1 interacts with at least 2 ligands: PD ligand–1 (PD-L1) and PD-L2 (21). In particular, the PD-1/PD-L1 pathway has been proposed as one of the most important mechanisms of T cell exhaustion and tolerance induction against infectious agents and tumors (19, 22–25). We therefore hypothesized that the PD-1/PD-L1 pathway plays a role in the loss of GVL effects in [B6→B6] mice. To test this hypothesis, we examined PD-1 expression on donor CD8⁺ T cells in lymph nodes on day +14 and +21 after BMT. It was significantly upregulated in allogeneic [B6→B6] recipients as compared with syngeneic controls but was low in [B6→C3] mice (Figure 4, A and B). We also investigated the expression of another inhibitory receptor, CTLA-4, on donor CD8⁺ T cells. Although the expression of cytoplasmic CTLA-4 was slightly upregulated in allogeneic animals as compared with syngeneic animals, its level did not differ between [B6→B6] and [B6→C3] mice (5.5% ± 1.0% vs. 4.5% ± 0.2%, respectively; P = 0.50).

Figure 4

Alloantigen expression on host non-hematopoietic cells stimulates PD-1 and its ligand pathway. [B6→B6] and [B6→C3] chimeras were transplanted as indicated in the legend for Figure 1 (n = 4–8). (A) Representative histogram of PD-1 expression among donor CD8⁺ T cells on day +14 and +21 in syngeneic (left), allogeneic [B6→B6] (middle), and [B6→C3] (right) recipients. (B) Frequencies of PD-1⁺ CD8⁺ T cells (mean ± SD). (C) Relative expressions of Pdl1 mRNA on day +14 and +21 in the livers of allogeneic [B6→B6] (gray bars) and allogeneic [B6→C3] mice (black bars). Data represent the mean (± SD) of n-fold difference in the amount of Pdl1 gene expression relative to that in syngeneic mice. (D) PD-L1 expression in the liver on day +14 (top row) and +21 (bottom row) from syngeneic (upper left) and allogeneic [B6→B6] (middle) and [B6→C3] (right) recipients. Isotype control of allogeneic [B6→B6] (lower left) is shown. Original magnification, ×200. *P < 0.05 compared with allogeneic controls.

We next examined PD-L1 expression in the liver by real-time PCR after BMT. PD-L1 expression was markedly upregulated in the liver of allogeneic controls as compared with syngeneic controls (Figure 4C). In allogeneic [B6→C3] mice, it was slightly upregulated on day +14 but not on day +21. Immunohistochemical analysis confirmed upregulated expression of PD-L1 in the liver of [B6→B6] mice, as previously reported (Figure 4D) (21, 26). These results showed that alloantigen expression on GVHD target epithelium is associated with upregulation of the PD-1/PD-L1 interactions between donor T cells and GVHD target tissue.

Blockade of the interaction between PD-1 and PD-L1 enhances GVL activity. We next examined whether blocking the PD-1/PD-L1 pathway could enhance GVL activity. [B6→C3] and [B6→B6] chimeras were reirradiated and injected with TCD BM cells and CD8⁺ T cells from C3 donors. Mice were i.p. injected with 500 μg of anti–PD-L1 mAb on day 0 and then with 200 μg on days +3, +6, +9, +12, +15, and +18 after BMT. In [B6→B6] recipients, injection of anti–PD-L1 mAbs significantly restored T cell functions on day +14, as assessed by CD69 expression (Figure 5A), IFN-γ production (Figure 5B), and CTL activity (Figure 5C). In [B6→C3] mice, it marginally upregulated CD69 expression, IFN-γ production, and CTL activity, although differences were not statistically significant (Figure 5, A, B, and D). As a consequence, anti–PD-L1 mAb administration significantly increased the severity of GVHD in [B6→B6] mice (Figure 5E) but not in [B6→C3] mice (Figure 5F). PD-L1 blockade also significantly augmented GVL activity in [B6→B6] recipients injected with EL4 cells on day 0 (Figure 5G, P < 0.05). It also delayed leukemia death in [B6→C3] mice, although the difference was not statistically significant (Figure 5H, P = 0.38). In controls, PD-L1 blockade did not affect leukemia mortality in TCD-BMT recipients (Figure 5H) or [B6→B6] recipients of syngeneic B6 CD8⁺ T cells (data not shown).

Figure 5

Blockade of the interaction between PD-1 and PD-L1 enhances GVL activity. [B6→C3] and [B6→B6] chimeras were reirradiated and injected with 5 × 10⁶ TCD BM cells alone or with 1 × 10⁶ CD8⁺ T cells from C3 donors. Mice were i.p. injected with 500 μg of anti PD-L1 mAbs or controls on day 0 and then 200 μg thereafter on days +3, +6, +9, +12, +15, and +18. Splenocytes were harvested on day +14 to determine the number of CD8⁺CD69⁺ T cells (A) and IFN-γ–producing CD8⁺ T cells (B) and CTL activity against EL4 targets (C and D). Results from a representative experiment of 2 similar experiments (means ± SD, n = 7–8/group). Mean clinical GVHD scores (±SEM) (E and F) after BMT are shown (n = 5–7/group). (G and H) Leukemia mortality after BMT in [B6→B6] and [B6→C3] chimeras injected with EL4 cells on day 0 (n = 4–11/group). Data from two similar experiments were combined. αPD-L1, anti–PD-L1 mAbs. *P < 0.05 compared with the corresponding controls.

Alloantigens are expressed in three major sites in HSCT recipients: APCs, GVHD target epithelium, and leukemia cells. Alloantigen expression on APCs is essential for the induction of GVHD (6), and an optimal GVL response occurs when alloantigens are expressed on both host APCs and tumor cells (7). Alloantigen expression on the epithelium is also critical for the induction of GVHD across mHA disparities (10), but GVHD can occur in the absence of alloantigen expression on epithelium in MHC-mismatched BMT (9). In this study, we addressed the effect of alloantigen expression on target epithelium in GVL using chimeric mouse models of GVHD and GVL across mHA disparities. Our models mimic clinical BMT in patients not in remission, since most of the mice relapsed after allogeneic BMT. This high tumor burden enabled us to compare the magnitude of GVL activity in our models, and we made surprising observations that alloantigen expression on non-hematopoietic cells inhibited GVL effects but enhanced GVHD. This observation challenges the current paradigm that GVL activity is strongly correlated with the severity of GVHD (1, 2, 27).

We found that alloantigen expression on non-hematopoietic cells induced donor T cell apoptosis and led to a contraction in the size of an alloreactive donor CD8⁺ T cell pool early after BMT. The remainder of the donor T cells were alive, but their ability to produce cytokines and cytotoxicity were impaired. This defect is similar to T cell exhaustion, which is a principal reason for the inability of the host to eliminate the persisting pathogen in chronic infections (18, 28). CD8⁺ T cell proliferation and differentiation into cytolytic effectors on an encounter with antigens are variable and change as a consequence of the antigen load (29). As the magnitude of the viral load increases, virus-specific T cells become more functionally impaired. During persistent infection, a high antigen load drives a significant number of virus-specific T cells into activation-induced apoptosis, and the remaining virus-specific T cells remain alive but in a dysfunctional state of cytotoxicity (18, 30–33). In tumor models, antigen quantity determines the behavior of the CD8⁺ effector cells, including their effector function and sensitivity to apoptosis (34–36). In patients with a larger tumor burden, CD8⁺ T cells were found to undergo apoptosis (37). Thus, a higher alloantigen load in allogeneic controls as compared with chimeras, in which alloantigen expression is limited to hematopoietic cells and tumor cells, may induce apoptosis and the dysfunction of alloreactive T cells, which leads to the inability of the host to eliminate leukemia.

Our results are consistent with seminal observations by Meunier, Fontaine, and colleagues, who showed that the adoptive transfer of immunodominant mHA (B6^dom1)–specific T cells eradicates B6^dom1-expressing leukemia more efficiently in mice lacking B6^dom1 expression than in mice expressing B6^dom1 (38). This was because the widespread expression of B6^dom1 caused activation-induced apoptosis and dysfunction of donor T cells in mice expressing B6^dom1 (38, 39). These findings along with our results indicate that allogeneic cellular therapy targeting mHAs exclusively expressed on APCs and tumor cells can induce a potent GVL effect while inducing less-severe GVHD than immunotherapy via targeting of ubiquitously expressed mHAs (40).

The PD-1/PD-L1 pathway is critically involved in T cell exhaustion and tolerance induction in infection and tumor immunology (18–20, 23–25, 41). It is also required for protection against chronic rejection of cardiac allograft, and induction of peripheral deletional tolerance of alloreactive, anti-donor CD8⁺ T cells to achieve successful engraftment in BMT (42, 43). In this study, we found that PD-1 expression was upregulated in donor T cells and PD-L1 expression was upregulated in GVHD target organs. The expression of PD-1/PD-L1 was markedly reduced in chimeras lacking alloantigen expression on non-hematopoietic cells. PD-1 and PD-L1 expression is induced upon cell activation and inflammation in GVHD (44); therefore, the absence of alloantigen expression on GVHD target epithelium reduced GVHD in chimeric mice, which resulted in insufficient stimulation of the PD-1/PD-L1 interaction. Target tissue expression of PD-L1 is also critical for the induction of T cell exhaustion or tolerance in chronic viral infection, autoimmune diabetes, and cardiac allografting (19, 42, 45).

Both PD-1 and PD-L1 were markedly upregulated in [B6→B6] mice, but they were also modestly upregulated in [B6→C3] mice. Blockade of PD-1/PD-L1 interactions significantly restored T cell effector functions in [B6→B6] mice but modestly restored them in [B6→C3] mice as well. The relevance of these observations is shown by the PD-1/PD-L1 blockade studies. These data showed that the PD-1/PD-L1 pathway is particularly germane to [B6→B6] mice with widespread expression of alloantigens but also applies, at least in part, to [B6→C3] mice, wherein alloantigen expression is only on APCs. While there is likely to be a role for this pathway in the absence of epithelial alloantigen expression, the full negative impact of this pathway on GVL is only seen when alloantigen expression is present on non-hematopoietic tissues.

Of note, the improvement in GVL by the PD-1/PD-L1 blockade was partial, as has been shown in chronic viral infection (46–48). This may be due to the presence of multiple negative regulatory pathways that contribute to T cell exhaustion, including CTLA-4, IL-10, LAG-3, CD160, and 2B4 (20, 47, 49). In addition, the population of exhausted T cells is heterogeneous, and this intervention is effective only for PD-1^lo and not PD-1^hi, which are subsets of exhausted T cells (50). Many of these inhibitory receptors are either coexpressed by the same exhausted T cells or differentially expressed on different subsets of exhausted cells. As the severity of the infection increases, the number of different inhibitors expressed per cell increases (47). A second inhibitory receptor, CTLA-4, can be overexpressed by exhausted CD4⁺ T cells in chronic viral infection, but it appears to have a minimal role on exhausted CD8⁺ T cells (19, 51). Although CTLA-4 was only slightly upregulated on CD8⁺ T cells in contrast to the marked upregulation of PD-1 in our CD8-dependent model of MHC-matched BMT, the precise inhibitory receptors of therapeutic interest may differ between CD4⁺-dependent and CD8⁺-dependent GVHD/GVL. Another key negative regulatory pathway is mediated by Foxp3⁺ Tregs. However, enhancement of GVL is not due to effects of the PD-1/PD-L1 blockade on Tregs, because blockade of PD-1/PD-L1 interactions enhances the expansion and function of Tregs (52). The hierarchy of these pathways in regulating GVL will need to be studied in the future based on better understanding of the delineation of T cell subsets and models (53). However, our results suggest the detrimental effect of GVHD-induced immunosuppression on GVL responses, regardless of which inhibitory pathway might be dominant clinically.

In addition, the administration of anti–PD-L1 mAb also exacerbated acute GVHD, as has been shown in a previous study (54). Therefore, the beneficial effects of the PD-1/PD-L1 blockade may be offset by the exacerbation of GVHD. Effects of the inhibitory receptor blockade might depend on the magnitude or stage of donor T cell activation and the severity of GVHD; therefore, the timing and duration of the targeting may be important.

In clinical HSCT, alloantigens continue to be presented on MHC class I in non-hematopoietic cells throughout the lifetime of the transplant recipients. However, a substantial number of patients eventually develop tolerance after resolution of GVHD and often experience leukemia relapse. Although activation-induced apoptosis of alloreactive T cells has been proposed as an explanation of this paradox (55), studies monitoring GVHD-specific T cell clones indicate that host-reactive T cells are continuously present after allogeneic HSCT (56–58). Our results provide a logical explanation for this paradox. However, the process of exhaustion is unlikely to occur in patients not developing GVHD, because induction of T cell exhaustion requires antigen-specific activation of T cells and subsequent differentiation into effector T cells. In these patients, tolerance could be induced by other mechanisms, such as functional central and peripheral tolerance mechanisms. It is well known that GVL is not apparent in patients with high leukemia burden. Although leukemia cells used in the current study do not express PD-L1 (22, 59), leukemia cells expressing PD-L1 may also directly limit the GVL response in patients with high leukemia burden (22, 24, 25). However, such insights from animal models must be extrapolated with caution to clinical studies involving humans.

It has been assumed that T cell exhaustion is antigen specific in chronic viral infection. Bystander lysis of T cells has also been reported in the course of viral infections (60), but is of minimal significance because of its limited magnitude and because normal thymic function can replenish the peripheral T cell pool. In contrast, in GVHD, T cell exhaustion occurs after initial T cell activation and the subsequent development of GVHD. GVHD induces bystander apoptosis of non-host-reactive T cells. In addition, GVHD-mediated injury of the thymus and the secondary lymphoid organs inhibits full replenishment of the peripheral T cell pool (55). Thus, establishment of full immune competence probably requires the additional process of T cell reconstitution following T cell exhaustion.

In conclusion, our results indicated the significance of alloantigen expression on non-hematopoietic cells in GVL. Alloantigen expression on non-hematopoietic cells induces the apoptosis of donor T cells and the dysfunction of cytotoxic effector function, which leads to a reduction in GVL activity. T cell dysfunction was partially restored by blocking PD-1/PD-L1 interactions, which suggests that the therapeutic “tuning” of T cell responses via modulation of negative regulatory pathways represents a novel strategy for enhancing GVL. Our results in combination with those of previous studies (6, 7, 9, 10, 38, 39) provide a complete picture of the effect of alloantigen expression on host APCs, GVHD target epithelium, and tumor cells in allogeneic HSCT; alloantigen expression on host non-hematopoietic cells augments GVHD but suppresses GVL effects. This concept may provide an important framework for understanding the pathophysiology of GVHD and allow for the separation of GVHD and GVL.

We thank Miyuki Azuma of Tokyo Medical and Dental University for providing hybridoma MIH5-producing anti–PD-L1 mAbs. This study was supported by grant 21390295 from the Ministry of Education, Culture, Sports, Science, and Technology (Tokyo, Japan) (to T. Teshima), Health and Labor Science Research Grants (Tokyo, Japan) (to T. Teshima), and a grant from the Foundation for Promotion of Cancer Research (Tokyo, Japan) (to T. Teshima).

Address correspondence to: Takanori Teshima, Center for Cellular and Molecular Medicine, Kyushu University Hospital, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan. Phone: 81.92.642.5947; Fax: 81.92.642.5951; E-mail: tteshima@cancer.med.kyushu-u.ac.jp.

Conflict of interest: The authors have declared that no conflict of interest.

Reference information: J Clin Invest. 2010;120(7):2370–2378. doi:10.1172/JCI39165.

1. Weiden P, et al. Antileukemic effect of graft-versus-host disease in human recipients of allogeneic-marrow grafts. N Engl J Med. 1979;300(19):1068–1073.
   View this article via: PubMed Google Scholar
2. Weiden PL, Sullivan KM, Flournoy N, Storb R, Thomas ED. Antileukemic effect of chronic graft-versus-host disease: contribution to improved survival after allogeneic marrow transplantation. N Engl J Med. 1981;304(25):1529–1533.
   View this article via: PubMed Google Scholar
3. Korngold R, Sprent J. Lethal graft-versus-host disease after bone marrow transplantation across minor histocompatibility barriers in mice. Prevention by removing mature T-cells from marrow. J Exp Med. 1978;148(6):1687–1698.
   View this article via: PubMed CrossRef Google Scholar
4. Apperley JF, Jones L, Hale G, Goldman JM. Bone marrow transplantation for chronic myeloid leukemia: T-cell depletion with Campath-1 reduces the incidence of acute graft-versus-host disease but may increase the risk of leukemia relapse. Bone Marrow Transplant. 1986;1(1):53–66.
   View this article via: PubMed Google Scholar
5. Atkinson K, et al. Risk factors for chronic graft-versus-host disease after HLA-identical sibling bone marrow transplantation. Blood. 1990;75(12):2459–2464.
   View this article via: PubMed Google Scholar
6. Shlomchik WD, et al. Prevention of graft versus host disease by inactivation of host antigen-presenting cells. Science. 1999;285(5426):412–415.
   View this article via: PubMed CrossRef Google Scholar
7. Reddy P, Maeda Y, Liu C, Krijanovski OI, Korngold R, Ferrara JL. A crucial role for antigen-presenting cells and alloantigen expression in graft-versus-leukemia responses. Nat Med. 2005;11(11):1244–1249.
   View this article via: PubMed CrossRef Google Scholar
8. Bleakley M, Riddell SR. Molecules and mechanisms of the graft-versus-leukaemia effect. Nat Rev Cancer. 2004;4(5):371–380.
   View this article via: PubMed CrossRef Google Scholar
9. Teshima T, et al. Acute graft-versus-host disease does not require alloantigen expression on host epithelium. Nat Med. 2002;8(6):575–581.
   View this article via: PubMed Google Scholar
10. Jones SC, Murphy GF, Friedman TM, Korngold R. Importance of minor histocompatibility antigen expression by nonhematopoietic tissues in a CD4+ T cell-mediated graft-versus-host disease model. J Clin Invest. 2003;112(12):1880–1886.
    View this article via: JCI PubMed Google Scholar
11. Ruggeri L, et al. Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants. Science. 2002;295(5562):2097–2100.
    View this article via: PubMed CrossRef Google Scholar
12. Matte CC, et al. Donor APCs are required for maximal GVHD but not for GVL. Nat Med. 2004;10(9):987–992.
    View this article via: PubMed Google Scholar
13. Cooke KR, et al. An experimental model of idiopathic pneumonia syndrome after bone marrow transplantation. I. The roles of minor H antigens and endotoxin. Blood. 1996;88(8):3230–3239.
    View this article via: PubMed Google Scholar
14. Korngold R, Sprent J. Features of T cells causing H-2-restricted lethal graft-vs.-host disease across minor histocompatibility barriers. J Exp Med. 1982;155(3):872–883.
    View this article via: PubMed CrossRef Google Scholar
15. Zhang Y, Joe G, Hexner E, Zhu J, Emerson SG. Alloreactive memory T cells are responsible for the persistence of graft-versus-host disease. J Immunol. 2005;174(5):3051–3058.
    View this article via: PubMed Google Scholar
16. Stefanova I, Dorfman JR, Germain RN. Self-recognition promotes the foreign antigen sensitivity of naive T lymphocytes. Nature. 2002;420(6914):429–434.
    View this article via: PubMed CrossRef Google Scholar
17. Zhang Y, Louboutin JP, Zhu J, Rivera AJ, Emerson SG. Preterminal host dendritic cells in irradiated mice prime CD8+ T cell-mediated acute graft-versus-host disease. J Clin Invest. 2002;109(10):1335–1344.
    View this article via: JCI PubMed Google Scholar
18. Zajac AJ, et al. Viral immune evasion due to persistence of activated T cells without effector function. J Exp Med. 1998;188(12):2205–2213.
    View this article via: PubMed CrossRef Google Scholar
19. Barber DL, et al. Restoring function in exhausted CD8 T cells during chronic viral infection. Nature. 2006;439(7077):682–687.
    View this article via: PubMed Google Scholar
20. Shin H, Wherry EJ. CD8 T cell dysfunction during chronic viral infection. Curr Opin Immunol. 2007;19(4):408–415.
    View this article via: PubMed Google Scholar
21. Keir ME, Butte MJ, Freeman GJ, Sharpe AH. PD-1 and its ligands in tolerance and immunity. Annu Rev Immunol. 2008;26:677–704.
    View this article via: PubMed CrossRef Google Scholar
22. Dong H, et al. Tumor-associated B7-H1 promotes T-cell apoptosis: a potential mechanism of immune evasion. Nat Med. 2002;8(8):793–800.
    View this article via: PubMed Google Scholar
23. Ahmadzadeh M, et al. Tumor antigen-specific CD8 T cells infiltrating the tumor express high levels of PD-1 and are functionally impaired. Blood. 2009;114(8):1537–1544.
    View this article via: PubMed Google Scholar
24. Zhang L, Gajewski TF, Kline J. PD-1/PD-L1 interactions inhibit anti-tumor immune responses in a murine acute myeloid leukemia model. Blood. 2009;114(8):1545–1552.
    View this article via: PubMed Google Scholar
25. Mumprecht S, Schurch C, Schwaller J, Solenthaler M, Ochsenbein AF. PD-1 signaling on chronic myeloid leukemia-specific T cells results in T cell exhaustion and disease progression. Blood. 2009;114(8):1528–1536.
    View this article via: PubMed Google Scholar
26. Iwai Y, Terawaki S, Ikegawa M, Okazaki T, Honjo T. PD-1 inhibits antiviral immunity at the effector phase in the liver. J Exp Med. 2003;198(1):39–50.
    View this article via: PubMed CrossRef Google Scholar
27. Horowitz MM, et al. Graft-versus-leukemia reactions after bone marrow transplantation. Blood. 1990;75(3):555–562.
    View this article via: PubMed Google Scholar
28. Gallimore A, et al. Induction and exhaustion of lymphocytic choriomeningitis virus-specific cytotoxic T lymphocytes visualized using soluble tetrameric major histocompatibility complex class I-peptide complexes. J Exp Med. 1998;187(9):1383–1393.
    View this article via: PubMed CrossRef Google Scholar
29. Welsh RM. Assessing CD8 T cell number and dysfunction in the presence of antigen. J Exp Med. 2001;193(5):F19–F22.
    View this article via: PubMed Google Scholar
30. Moskophidis D, Lechner F, Pircher H, Zinkernagel RM. Virus persistence in acutely infected immunocompetent mice by exhaustion of antiviral cytotoxic effector T cells. Nature. 1993;362(6422):758–761.
    View this article via: PubMed CrossRef Google Scholar
31. Appay V, et al. HIV-specific CD8(+) T cells produce antiviral cytokines but are impaired in cytolytic function. J Exp Med. 2000;192(1):63–75.
    View this article via: PubMed CrossRef Google Scholar
32. Xiong Y, et al. Simian immunodeficiency virus (SIV) infection of a rhesus macaque induces SIV-specific CD8(+) T cells with a defect in effector function that is reversible on extended interleukin-2 incubation. J Virol. 2001;75(6):3028–3033.
    View this article via: PubMed CrossRef Google Scholar
33. Pantaleo G, Harari A. Functional signatures in antiviral T-cell immunity for monitoring virus-associated diseases. Nat Rev Immunol. 2006;6(5):417–423.
    View this article via: PubMed CrossRef Google Scholar
34. Tham EL, Mescher MF. Signaling alterations in activation-induced nonresponsive CD8 T cells. J Immunol. 2001;167(4):2040–2048.
    View this article via: PubMed Google Scholar
35. Tham EL, Shrikant P, Mescher MF. Activation-induced nonresponsiveness: a Th-dependent regulatory checkpoint in the CTL response. J Immunol. 2002;168(3):1190–1197.
    View this article via: PubMed Google Scholar
36. Boissonnas A, et al. Antigen distribution drives programmed antitumor CD8 cell migration and determines its efficiency. J Immunol. 2004;173(1):222–229.
    View this article via: PubMed Google Scholar
37. Saito T, Dworacki G, Gooding W, Lotze MT, Whiteside TL. Spontaneous apoptosis of CD8+ T lymphocytes in peripheral blood of patients with advanced melanoma. Clin Cancer Res. 2000;6(4):1351–1364.
    View this article via: PubMed Google Scholar
38. Fontaine P, Roy-Proulx G, Knafo L, Baron C, Roy DC, Perreault C. Adoptive transfer of minor histocompatibility antigen-specific T lymphocytes eradicates leukemia cells without causing graft-versus-host disease. Nat Med. 2001;7(7):789–794.
    View this article via: PubMed CrossRef Google Scholar
39. Meunier MC, Roy-Proulx G, Labrecque N, Perreault C. Tissue distribution of target antigen has a decisive influence on the outcome of adoptive cancer immunotherapy. Blood. 2003;101(2):766–770.
    View this article via: PubMed CrossRef Google Scholar
40. Dickinson AM, et al. In situ dissection of the graft-versus-host activities of cytotoxic T cells specific for minor histocompatibility antigens. Nat Med. 2002;8(4):410–414.
    View this article via: PubMed Google Scholar
41. Ding ZC, Blazar BR, Mellor AL, Munn DH, Zhou G. Chemotherapy rescues tumor-driven aberrant CD4+ T-cell differentiation and restores an activated polyfunctional helper phenotype. Blood. 2010;115(12):2397–2406.
    View this article via: PubMed Google Scholar
42. Tanaka K, et al. PDL1 is required for peripheral transplantation tolerance and protection from chronic allograft rejection. J Immunol. 2007;179(8):5204–5210.
    View this article via: PubMed Google Scholar
43. Haspot F, et al. Peripheral deletional tolerance of alloreactive CD8 but not CD4 T cells is dependent on the PD-1/PD-L1 pathway. Blood. 2008;112(5):2149–2155.
    View this article via: PubMed Google Scholar
44. Yamazaki T, et al. Expression of programmed death 1 ligands by murine T cells and APC. J Immunol. 2002;169(10):5538–5545.
    View this article via: PubMed Google Scholar
45. Keir ME, et al. Tissue expression of PD-L1 mediates peripheral T cell tolerance. J Exp Med. 2006;203(4):883–895.
    View this article via: PubMed Google Scholar
46. Crawford A, Wherry EJ. The diversity of costimulatory and inhibitory receptor pathways and the regulation of antiviral T cell responses. Curr Opin Immunol. 2009;21(2):179–186.
    View this article via: PubMed CrossRef Google Scholar
47. Blackburn SD, et al. Coregulation of CD8+ T cell exhaustion by multiple inhibitory receptors during chronic viral infection. Nat Immunol. 2009;10(1):29–37.
    View this article via: PubMed Google Scholar
48. Petrovas C, et al. PD-1 is a regulator of virus-specific CD8+ T cell survival in HIV infection. J Exp Med. 2006;203(10):2281–2292.
    View this article via: PubMed CrossRef Google Scholar
49. Brooks DG, Trifilo MJ, Edelmann KH, Teyton L, McGavern DB, Oldstone MB. Interleukin-10 determines viral clearance or persistence in vivo. Nat Med. 2006;12(11):1301–1309.
    View this article via: PubMed CrossRef Google Scholar
50. Blackburn SD, Shin H, Freeman GJ, Wherry EJ. Selective expansion of a subset of exhausted CD8 T cells by alphaPD-L1 blockade. Proc Natl Acad Sci U S A. 2008;105(39):15016–15021.
    View this article via: PubMed CrossRef Google Scholar
51. Kaufmann DE, et al. Upregulation of CTLA-4 by HIV-specific CD4+ T cells correlates with disease progression and defines a reversible immune dysfunction. Nat Immunol. 2007;8(11):1246–1254.
    View this article via: PubMed CrossRef Google Scholar
52. Franceschini D, et al. PD-L1 negatively regulates CD4+CD25+Foxp3+ Tregs by limiting STAT-5 phosphorylation in patients chronically infected with HCV. J Clin Invest. 2009;119(3):551–564.
    View this article via: JCI PubMed CrossRef Google Scholar
53. Socie G, Blazar BR. Acute graft-versus-host disease: from the bench to the bedside. Blood. 2009;114(20):4327–4336.
    View this article via: PubMed Google Scholar
54. Blazar BR, et al. Blockade of programmed death-1 engagement accelerates graft-versus-host disease lethality by an IFN-gamma-dependent mechanism. J Immunol. 2003;171(3):1272–1277.
    View this article via: PubMed Google Scholar
55. Brochu S, Rioux-Masse B, Roy J, Roy DC, Perreault C. Massive activation-induced cell death of alloreactive T cells with apoptosis of bystander postthymic T cells prevents immune reconstitution in mice with graft-versus-host disease. Blood. 1999;94(2):390–400.
    View this article via: PubMed Google Scholar
56. Dey B, et al. The fate of donor T-cell receptor transgenic T cells with known host antigen specificity in a graft-versus-host disease model. Transplantation. 1999;68(1):141–149.
    View this article via: PubMed Google Scholar
57. Choi EY, et al. Real-time T-cell profiling identifies H60 as a major minor histocompatibility antigen in murine graft-versus-host disease. Blood. 2002;100(13):4259–4265.
    View this article via: PubMed Google Scholar
58. Michalek J, Collins RH, Hill BJ, Brenchley JM, Douek DC. Identification and monitoring of graft-versus-host specific T-cell clone in stem cell transplantation. Lancet. 2003;361(9364):1183–1185.
    View this article via: PubMed CrossRef Google Scholar
59. Hirano F, et al. Blockade of B7-H1 and PD-1 by monoclonal antibodies potentiates cancer therapeutic immunity. Cancer Res. 2005;65(3):1089–1096.
    View this article via: PubMed Google Scholar
60. Ando K, et al. Perforin, Fas/Fas ligand, and TNF-alpha pathways as specific and bystander killing mechanisms of hepatitis C virus-specific human CTL. J Immunol. 1997;158(11):5283–5291.
    View this article via: PubMed Google Scholar
61. Tsushima F, et al. Preferential contribution of B7-H1 to programmed death-1-mediated regulation of hapten-specific allergic inflammatory responses. Eur J Immunol. 2003;33(10):2773–2782.
    View this article via: PubMed CrossRef Google Scholar
62. Teshima T, et al. IL-11 separates graft-versus-leukemia effects from graft-versus-host disease after bone marrow transplantation. J Clin Invest. 1999;104(3):317–325.
    View this article via: JCI PubMed CrossRef Google Scholar